animal intelligence experimental studies

1st Edition

Animal Intelligence Experimental Studies

• Taylor & Francis eBooks (Institutional Purchase) Opens in new tab or window

Description

Animal Intelligence is a consolidated record of Edward L. Thorndike's theoretical and empirical contributions to the comparative psychology of learning. Thorndike's approach is systematic and comprehensive experimentation using a variety of animals and tasks, all within a laboratory setting. When this book first appeared, it set a compelling example, and helped make the study of animal behavior very much an experimental laboratory science.

This landmark study in the investigation of animal intelligence illustrates Thorndike's thinking on the evolution of the mind. It includes his formal statement of the influential law of effect, which had a significant impact on other behaviorists. Hull's law of primary reinforcement was closely related to the law of effect and Skinner acknowledged that the process of operant conditioning was probably that described in the law of effect.

The new introduction by Darryl Bruce is an in-depth study of Thorndike's legacy to comparative psychology as well as a thorough retrospective review of Animal Intelligence . He includes a biographical introduction of the behaviorist and then delves into his theories and work. Among the topics Bruce covers with respect to Thorndike's studies are the nature of animal intelligence, the laws of learning and connectionism, implications for comparative psychology, and relation to theories of other behaviorists. Animal Intelligence is an intriguing analysis that will be of importance to psychologists and animal behaviorists.

Table of Contents

Edward Thorndike

About VitalSource eBooks

VitalSource is a leading provider of eBooks.

• Access your materials anywhere, at anytime.
• Customer preferences like text size, font type, page color and more.
• Take annotations in line as you read.

Multiple eBook Copies

This eBook is already in your shopping cart. If you would like to replace it with a different purchasing option please remove the current eBook option from your cart.

Book Preview

The country you have selected will result in the following:

• Product pricing will be adjusted to match the corresponding currency.
• The title Perception will be removed from your cart because it is not available in this region.

• Project Gutenberg
• 74,202 free eBooks
• 3 by Edward L. Thorndike

Animal intelligence: Experimental studies by Edward L. Thorndike

Read now or download (free!)

Similar Books

About this ebook.

• Privacy policy
• About Project Gutenberg
• Terms of Use
• Contact Information

Breadcrumbs Section. Click here to navigate to respective pages.

Animal Intelligence

DOI link for Animal Intelligence

Get Citation

Animal Intelligence is a consolidated record of Edward L. Thorndike's theoretical and empirical contributions to the comparative psychology of learning. Thorndike's approach is systematic and comprehensive experimentation using a variety of animals and tasks, all within a laboratory setting. When this book first appeared, it set a compelling example, and helped make the study of animal behavior very much an experimental laboratory science.

This landmark study in the investigation of animal intelligence illustrates Thorndike's thinking on the evolution of the mind. It includes his formal statement of the influential law of effect, which had a significant impact on other behaviorists. Hull's law of primary reinforcement was closely related to the law of effect and Skinner acknowledged that the process of operant conditioning was probably that described in the law of effect.

The new introduction by Darryl Bruce is an in-depth study of Thorndike's legacy to comparative psychology as well as a thorough retrospective review of Animal Intelligence . He includes a biographical introduction of the behaviorist and then delves into his theories and work. Among the topics Bruce covers with respect to Thorndike's studies are the nature of animal intelligence, the laws of learning and connectionism, implications for comparative psychology, and relation to theories of other behaviorists. Animal Intelligence is an intriguing analysis that will be of importance to psychologists and animal behaviorists.

TABLE OF CONTENTS

Chapter 1 | 19  pages, the study of consciousness and the study of behavior, chapter 2 | 136  pages, animal intelligence; an experimental study of the associative processes in animals 1, chapter 3 | 13  pages, the instinctive reactions of young chicks 1, chapter 4 | 3  pages, a note on the psychology of fishes 1, chapter 5 | 69  pages, the mental life of the monkeys; an experimental study 1, chapter 6 | 41  pages, laws and hypotheses for behavior laws of behavior in general, chapter 7 | 13  pages, the evolution of the human intellect 1.

• Privacy Policy
• Terms & Conditions
• Cookie Policy
• Taylor & Francis Online
• Taylor & Francis Group
• Students/Researchers
• Librarians/Institutions

Connect with us

Registered in England & Wales No. 3099067 5 Howick Place | London | SW1P 1WG © 2024 Informa UK Limited

• DOI: 10.2307/1412682
• Corpus ID: 260440571

Animal Intelligence: An Experimental Study of the Associative Processes in Animals

• L. W. Kline , E. Thorndike
• American Journal of Psychology

567 Citations

In defense of ptc.

• Highly Influenced

Smolensky's proper treatment of connectionism: Having it both ways

An operating principle of the cerebral cortex, and a cellular mechanism for attentional trial-and-error pattern learning and useful classification extraction, feeling the future: a meta-analysis of 90 experiments on the anomalous anticipation of random future events, neural processes involved in action selection during a mixed-strategy game, an experimental study of belief learning using elicited beliefs, secondary teachers’ perceptions of the instructional evaluation process, the influence of ratio-reinforcement on video-gaming behaviour, post-stroke language remediation through constraint-induced aphasia therapy, principles of learning: a conceptual framework for domain-specific theories of learning.

• 12 Excerpts

Related Papers

Showing 1 through 3 of 0 Related Papers

Loading metrics

Open Access

Essays articulate a specific perspective on a topic of broad interest to scientists.

See all article types »

Leveraging social media and other online data to study animal behavior

Roles Conceptualization, Data curation, Visualization, Writing – original draft, Writing – review & editing

* E-mail: [email protected]

Affiliations School of Geography and the Environment, University of Oxford, Oxford, United Kingdom, WildCRU, Recanati-Kaplan Centre, Department of Zoology, Oxford University, Oxford, United Kingdom, School of Zoology, Tel Aviv University, Tel Aviv, Israel

Roles Conceptualization, Data curation, Writing – review & editing

Affiliations CEABN, Centro de Ecologia Aplicada Prof. Baeta Neves, InBIO Laboratório Associado, Instituto Superior de Agronomia, Universidade de Lisboa, Lisboa, Portugal, Helsinki Lab of Interdisciplinary Conservation Science (HELICS), Department of Geosciences and Geography, University of Helsinki, Helsinki, Finland

Affiliations Department of Anatomy, Cell Biology and Zoology, Faculty of Sciences, University of Extremadura, Badajoz, Spain, Ecology in the Anthropocene, Associated Unit CSIC-UEX, Faculty of Sciences, University of Extremadura, Badajoz, Spain

Roles Data curation, Writing – review & editing

Affiliation Poznań University of Life Sciences, Department of Zoology, Poznań, Poland

Affiliation Institute of Evolutionary Biology, Faculty of Biology, Biological and Chemical Research Centre, University of Warsaw, Warsaw, Poland

Affiliations Faculty of Environmental Sciences, Czech University of Life Sciences Prague, Prague, Czech Republic, TUM School of Life Sciences, Ecoclimatology, Technical University of Munich, Freising, Germany, Institute for Advanced Study, Technical University of Munich, Garching, Germany

Affiliation Mitrani Department of Desert Ecology, Jacob Blaustein Institutes of Desert Research, Ben-Gurion University of the Negev, Ben-Gurion, Israel

Affiliation Department of Ecology and Evolutionary Biology, University of California, Los Angeles, California, United States of America

Roles Data curation, Visualization, Writing – review & editing

Affiliations Université Paris-Saclay, CNRS, AgroParisTech, Ecologie Systématique Evolution, Gif sur Yvette, France, Biology Centre of the Czech Academy of Sciences, Institute of Hydrobiology, České Budějovice, Czech Republic

Roles Conceptualization, Data curation, Visualization, Writing – review & editing

Affiliation Department of Marine Renewable Resources, Institute of Marine Science (ICM-CSIC), Barcelona, Spain

• Reut Vardi, 
• Andrea Soriano-Redondo, 
• Jorge S. Gutiérrez, 
• Łukasz Dylewski, 
• Zuzanna Jagiello, 
• Peter Mikula, 
• Oded Berger-Tal, 
• Daniel T. Blumstein, 
• Ivan Jarić, 
• Valerio Sbragaglia

Published: August 29, 2024

• https://doi.org/10.1371/journal.pbio.3002793
• Reader Comments

This is an uncorrected proof.

The widespread sharing of information on the Internet has given rise to ecological studies that use data from digital sources including digitized museum records and social media posts. Most of these studies have focused on understanding species occurrences and distributions. In this essay, we argue that data from digital sources also offer many opportunities to study animal behavior including long-term and large-scale comparisons within and between species. Following Nikko Tinbergen’s classical roadmap for behavioral investigation, we show how using videos, photos, text, and audio posted on social media and other digital platforms can shed new light on known behaviors, particularly in a changing world, and lead to the discovery of new ones.

Citation: Vardi R, Soriano-Redondo A, Gutiérrez JS, Dylewski Ł, Jagiello Z, Mikula P, et al. (2024) Leveraging social media and other online data to study animal behavior. PLoS Biol 22(8): e3002793. https://doi.org/10.1371/journal.pbio.3002793

Copyright: © 2024 Vardi et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: R.V. was partly funded by the Alexander and Eva Lester post-doctoral fellowship. A.S-R. was supported by grant 2022.01951.CEECIND from the Portuguese Foundation for Science and Technology. P.M. was supported by the Faculty of Environmental Sciences CZU Prague within the framework of the Research Excellence in Environmental Sciences (REES 003) and by IAS TUM – Hans Fisher Senior Fellowship. I.J. was supported by grant no. 23-07278S from the Czech Science Foundation. V.S. is supported by a Ramón y Cajal research fellowship (RYC2021-033065-I) granted by the Spanish Ministry of Science and Innovation and he also acknowledges the Spanish government through the ‘Severo Ochoa Centre of Excellence’ accreditation (CEX2019-000928-S). Funders played no rule in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Rapidly accumulating digital data offer numerous opportunities for science. With more than half of the world’s population online ( https://www.itu.int/en/ITU-D/Statistics/Pages/stat/default.aspx ), billions of people are generating online digital data in the form of text, images, videos, and audio uploaded to social media platforms and other websites ( Box 1 ). Furthermore, field notes, printed books, and old news media are being increasingly digitized and made available online [ 1 ]. These vast digital knowledge repositories can provide meaningful insights into the natural world. Indeed, several emerging fields have been developed for that purpose; conservation culturomics uses digital data to inform conservation science and human–nature interactions [ 2 ], while iEcology (or passive crowdsourcing [ 3 ]) uses such data to study ecological patterns [ 4 ]. Indeed, geotagged data from multiple digital sources can complement other data to monitor distributions and occurrences of species, particularly of charismatic ones, or in and around human-dominated landscapes such as urban habitats or areas subjected to high human visitation [ 5 , 6 ].

Box 1. Categories of digital data

While using the term digital data, we distinguish between 3 major categories:

• Digitized scientific databases, such as digitized museum records, and audio or video online libraries, that have usually been collected by researchers.
• Citizen/community science data sets where members of the public record their nature sightings for scientific use, either for general data repositories or for specific research projects (e.g., iNaturalist and eBird).
• Social media platforms—such as X (formerly known as Twitter), Instagram, or Google Images—where individuals upload content generated for various purposes typically not with the intention to address scientific questions yet may, nevertheless, be relevant to research.

Data from the 3 categories can differ in their collection protocols, reliability, accuracy, accompanied metadata, and data-sharing rights. While we consider the importance of data use from all 3 categories, given the novelty, extent, and challenges associated with using data from social media platforms, we focus primarily on the potential and limitations of such digital data sources.

Digital data can also be used to characterize animal behavior [ 7 ]. For example, Jagiello and colleagues [ 8 ] used YouTube videos to compare the occurrence of various behaviors of Eurasian red squirrels and invasive gray squirrels ( Sciurus vulgaris and S . carolinensis ) between 2 habitats. They found that calling and aggressive behaviors were more frequent in forests than in urban habitats ( Fig 1 ). Similarly, Boydston and colleagues [ 9 ] analyzed YouTube videos to understand the structure and putative function of coyote–dog ( Canis latrans – C . familiaris ) interactions. They found evidence of intricate social behavior between the 2 species. However, YouTube is not the only platform that offers data that, while collected for other purposes, can be meaningful for behavioral ecology. Other sources may include various social media platforms (X (formerly Twitter), Facebook, Instagram, etc.), digitized scientific records, and citizen science databases (see Box 1 ). Such alternative sources of information may help fill important gaps in our understanding of animal behavior and shed light on how animal behavior may be influenced by humans’ actions.

• PPT PowerPoint slide
• PNG larger image
• TIFF original image

(A) Digital data (inner circle; photos, videos, and audio) can complement experimental and observational approaches aiming to characterize several aspects of animal behavior, such as social interactions and biological rhythms (middle circle). Applications of digital data are particularly interesting for characterizing behavioral and ecological patterns addressing several research fields (e.g., urban ecology and biological invasions) as well as tackling conservation issues (outer circle). ( B–D) Representative examples of studies that used digital data to characterize animal behavior. ( B ) Percentage of recorded behavior in forest and urban ecosystems for the European red squirrel ( Sciurus vulgaris ) based on YouTube videos (right; adapted from [ 8 ]); photo of a red squirrel (photo credit: Peter Mikula); ( C ) Density maps showing the distribution of bat predation records by diurnal birds based on published literature (left map) and online records such as Google images, Flickr, and YouTube (right map; adapted from [ 10 ], countries borders map taken from https://public.opendatasoft.com/explore/dataset/ne_10m_admin_0_countries/map/ ). Example photo of a European bee-eater ( Merops apiaster ) trying to swallow a Kuhl’s pipistrelle bat ( Pipistrellus kuhlii ; photo credit: Shuki Cheled). ( D ) Wilson’s phalarope ( Phalaropus tricolor ) spinning (counterclockwise) in tight circles to upwell small prey and feed upon them as revealed by freely available videos on YouTube, Vimeo, and Flickr (photo credit: Miroslav Šálek). Nearest neighbors are more likely to spin in the same direction, thus reducing interference with each other (adapted from [ 11 ]).

https://doi.org/10.1371/journal.pbio.3002793.g001

In the mid-20th century, Nikko Tinbergen created a foundational framework for the integrative study of animal behavior [ 12 , 13 ] by posing 4 interlinked questions regarding the 4 main axes of behavior: causation , the mechanistic basis of behavior; ontogeny , its development throughout an individual’s lifetime; evolution , its changes over an evolutionary time scale; and function , its adaptive value and current utility. Answering Tinbergen’s questions can be hindered by many research challenges including, but not restricted to, limited funds, time, accessibility, and sample sizes. In such cases, readily available data from various online platforms such as citizen science databases or social media platforms (for example, YouTube, Facebook, or Flickr) can prove to be a powerful and complementary tool to traditional methods involving observations and experiments ( Fig 2 ) [ 4 , 7 ]. Furthermore, social media platforms, similar to citizen science platforms, can also provide bridges between scientists and nature enthusiasts (as well as the general public) that can be harnessed to help create and review large data sets. This, in turn, can also encourage people to reconnect with nature and promote biodiversity conservation [ 14 ].

Traditionally, animal behavior has been studied mostly with empirical approaches and literature surveys. The addition of digital data enables us to explore ecological patterns (iEcology) and human–nature interactions (conservation culturomics). All of these approaches can help address Tinbergen’s questions of behavior. In return, Tinbergen’s questions help direct and shape research questions, experimental setups, and data collection. Conservation culturomics infers human behavior related to nature and is thus represented with a dashed arrow. Icons taken from https://openclipart.org/ .

https://doi.org/10.1371/journal.pbio.3002793.g002

Here, we propose that digital data, especially from social media platforms, can be used to answer questions beyond species distribution and occurrence to advance the field of animal behavior ( Fig 2 ). While keeping in mind that Tinbergen’s questions are interlinked and complementary to each other, we explore each question separately, highlighting both opportunities and challenges in using digital data to answer them. We further highlight the increased relevance of Tinbergen’s questions to biodiversity conservation. We showcase instances where digital data has already been used to study animal behavior ( Fig 1 and S1 Table ) and suggest possible avenues for further research incorporating digital data to address fundamental and applied behavioral issues.

Studies dealing with causation try to understand what causes a behavior to be performed. When combined with remotely sensed, freely available data, digital data sources can be used to explore the external mechanisms underlying a behavioral trait. For example, Cabello-Vergel and colleagues [ 15 ] combined data on the thermoregulatory behavior of individual storks (Ciconiidae) from georeferenced images and videos found at the Macaulay Library repository ( https://www.macaulaylibrary.org ) with remotely sensed microclimate data. They investigated the determinants of “urohidrosis” (excreting onto the legs as a form of evaporative cooling) in 19 stork species. They found that high heat loads (high temperature, humidity, and solar radiation, and low wind speed) promoted the use of urohidrosis and thus evaporative heat loss. In the face of global climate change, exploring shifts in mechanisms of control with microclimate data can inform us about mechanisms of adaptation to changing environments and provide profound insights facilitating future conservation efforts.

The study of social learning and the emergence of novel and innovative behaviors in relation to environmental conditions could particularly benefit from digital data sources because people often record surprising or unexpected animal behaviors [ 7 ]. For example, data from multiple digital data sources revealed that 10 out of the 16 world’s terrestrial hermit crab species ( Coenobitidae ) widely use artificial shells, predominantly plastic caps, but also pieces of glass or metal [ 16 ]. This novel behavior may be driven by decreased availability of gastropod shells, sexual signaling, lightness of artificial shells, odor cues, and/or camouflage in a polluted environment. Together with controlled preferences experiments and/or records of pollution levels and other environmental conditions, we can address the underlying mechanisms of this behavior, which may ultimately influence the evolutionary trajectory of the species. Other examples include YouTube videos that have been used to describe horses opening doors and gate mechanisms [ 17 ] or investigate death-related behavioral responses in Asian elephants ( Elephas maximus ) such as carrying dead calves [ 18 ]. Understanding why and when these rare behaviors occur may not be possible without such online records.

In 2022, Møller and Xia [ 19 ] showed that bird species recorded on YouTube videos feeding directly from people’s hands also presented more innovative behaviors, had a higher rate of introduction success, and greater urban tolerance than species not recorded displaying such behavior. This demonstrates the connections between Tinbergen’s questions and highlights that an individual’s (or species) ability to respond behaviorally to external conditions may also rely on its evolutionary history and affects its chances of survival. It further shows that the fields of urban ecology and invasion biology can greatly benefit from integrating these novel digital data sources. For example, with most of the global human population living in cities and the omnipresence of online social platforms, digital data can make global multi-city comparisons of urbanization effects on species behavior feasible. Moreover, human activity can be easily tracked using mobility reports provided by Google ( https://www.google.com/covid19/mobility/ ) and Apple ( https://covid19.apple.com/mobility ). These can provide a high-resolution understanding of where and when humans are active and how they can play an important role in shaping animal behavior. Such knowledge can help enhance studies of antipredator behavior and wildlife tolerance, as it was used to study the consequences of the COVID-19 pandemic lockdowns [ 20 , 21 ]. Likewise, documenting first arrivals and monitoring the spread of invasive species, their behavior, and interactions with native species can become more efficient by incorporating digital data from online repositories [ 22 ].

We acknowledge that digital sources alone cannot offer many insights into internal mechanisms of behavior, such as hunger state or past experience (exceptions may include behaviors that are influenced by temperature, which may be inferred if the data are georeferenced and time stamped). Studying proximate physiological mechanisms often requires extensive field and laboratory experiments. However, addressing what mechanisms drive behavior in terms of changes in the external stimulus (social and physical environment) could greatly benefit from the copious number of available images and videos online. This is particularly true considering current and future global environmental challenges.

Digital data in the forms of images, audio, videos, and live-streaming videos can also be used to study and quantify different behavioral shifts in individuals over their lifetimes. For example, using online-sourced photographs, Naude and colleagues [ 23 ] showed that adult martial eagles ( Polemaetus bellicosus ) preyed more on birds than juveniles and subadults, which preferred less agile reptiles and mammals. They attributed this pattern to an improvement in hunting skills with age. Another study found evidence for “ontogenetic deepening”—the phenomenon that older and larger fish are found in deeper water, whereas younger and smaller fish stay in shallower water—in dusky groupers ( Epinephelus marginatus ) using YouTube videos of recreational fishers [ 24 ]. Exploring videos over several years, they further showed that fishing depth did not change over time and thus suggested that this ontogenetic deepening may not be solely driven by changes in harvesting pressure. Combining acoustic recordings from various sources (field recordings, a museum sound library, and citizen science records), Riós-Chelén and colleagues [ 25 ] found that birds can adapt their songs to environmental acoustic conditions. The fact that songbirds (known as oscines), who learn their songs, showed stronger associations between environmental noise and song modifications than other closely related bird species with innate songs (suboscines) indicates the involvement of ontogenetic processes in this adjustment.

Other studies can use similar approaches to further explore ontogenetic changes in different species’ hunting skills, aggressiveness, mating rituals, and parental care, with or without complementing intensive fieldwork (see S1 Table ). Exploring such changes in behavior in response to anthropogenic environmental changes worldwide can be of great importance for conservation science, urban ecology, and agroecology. For example, live-streaming videos of bird nests—which have become very common for many species and sites (e.g., https://camstreamer.com/blog/streaming-birds-with-an-eagle-eye and https://www.viewbirds.com/ )—can provide rich information to study the development of nestling vocal signals, the learning of songs, or the establishment of siblings relationships, as well as differences in such behaviors as a function of the distance to urban areas, human disturbance level, or levels of noise or light pollution [ 26 ]. Nonetheless, similar to exploring causation mechanisms, answering questions related to ontogeny cannot solely rely on digital data sources since ontogenetic processes often involve studying individuals over time. Furthermore, developing a deep understanding of external factors affecting the development of behavior may also require well-designed controlled experiments, which can be more challenging to accomplish with the available digital databases.

With images and videos from around the world spanning several decades available online, it is now possible to use digital data to explore intra- and interspecific traits and behaviors, as well as study their evolution in the light of anthropogenic environmental changes. For example, using crowd-sourced images and videos, Mikula and colleagues [ 10 ] showed that predator–prey interactions between diurnal birds and bats, which were previously thought to be rare, have been commonly reported around the world ( Fig 1 ). This indicates that diurnal bird predation might act as one of the drivers of the evolution of bat nocturnality. Similarly, using social media videos and phylogenetic modeling, Bastos and colleagues [ 27 ] showed that tool-using behavior in parrots is far more common than previously thought and that these new sources of data can be used to better understand the origin, evolution, and drivers of rare behaviors. In another example, Pearse and colleagues [ 28 ] were able to explore evolutionary patterns in bird song at a broad scale (in terms of pitch and complexity) using a large citizen science digital repository, combined with scientific data on bird biology, life history, and geographical distribution, and advanced machine learning techniques. Surprisingly, they showed that suboscine and oscine birds have similar song complexity. They further noted that using Artificial intelligence (AI) tools to help analyze citizen science data can further facilitate research on bird song evolution. However, such tools may also have limitations and need to be routinely validated and assessed.

The fact that digital repositories can potentially hold decades-old data allows retrospective explorations of data collected long before the research has commenced. For example, the COVID-19 pandemic highlighted the importance and usefulness of citizen science data sets, as past records could be compared with records under the novel environmental settings created by the pandemic [ 21 ]. Similar data sets may be obtained from various social media platforms that are far more popular than citizen science platforms, both in volume and in geographic coverage. For example, there are 3 million iNaturalist users ( https://www.inaturalist.org/stats ) compared with 300 million X (formerly Twitter) users ( https://www.statista.com/statistics/303681/twitter-users-worldwide/ ). While most of the content on X would probably be irrelevant for ecology and conservation, the potential to reach and engage new audiences, and access diverse data could be valuable. Using these novel data sources can further facilitate large spatial scale explorations of evolutionary changes in animal behavior. It may also help researchers to better plan and choose field sites before embarking on intensive fieldwork.

Many aspects of the evolution of animal behavior are challenging to document directly because numerous phenotypic traits co-evolve over large spatial and phylogenetic scales, making comparative studies useful. For example, body coloration may be an important factor in answering fundamental questions in behavioral ecology that provides insights into local behavioral adaptations [ 29 , 30 ]. Online image repositories have already been used to document geographical and phylogenetic variation in color patterns in birds and mammals, including color polymorphism [ 31 ], mutations [ 32 ], and variation in the morphology of color strips and patches [ 33 ]. In addition to readily available data, people can be encouraged to upload their images, videos, and sound recordings for specific studies through citizen science platforms [ 34 ] or social media platforms [ 35 ]. Spatial data on the phenotypic distributions are often collected via field observations and inspection of voucher specimens.

We envision that online images, videos, and acoustic recordings may provide a rich resource of information on large-scale variation in many phenotypic traits closely linked to animal behavior, such as nest morphology in fish and birds, or the size and shape of ornaments and armaments (e.g., antlers in deer or bony spurs in birds). Yet, we must acknowledge the limitations of using digital data to answer questions of an evolutionary nature that require some genomic knowledge. Still, the sheer volume of digital data and the ability to compare data of many species and populations inhabiting different areas and environments can provide valuable information for the processes and mechanisms involved in evolutionary adaptation and speciation.

Answering function-related questions—how a behavior increases one’s fitness through survival and reproduction—can also gain much from using digital data. With the ubiquity of the Internet, we can explore external drivers of current utility and sexual selection regarding behavioral contributions to overall fitness. These may include intra- and interspecific interactions, migratory patterns, predation risk, and mating rituals. For example, using live-streaming underwater cameras, Coleman and Burge [ 36 ] showed a higher association between sand tiger sharks ( Carcharias taurus ) and round scads ( Decapterus punctatus ) in the presence of scad mesopredators, which enhances foraging opportunities for sand tiger sharks and reduces predation risk for the scads. Such behaviorally mediated indirect interactions may have far-reaching implications for trophic interactions, including predator and prey strategies. Studies like this highlight the potential of these novel data and technologies in ecological research.

Digital data can be further used to study the timing of biological processes (i.e., phenology) in animals and how these are being affected by external cues such as climate change, land use changes, or human disturbance. For example, using Wikipedia page views, Mittermeier and colleagues [ 37 ] tracked seasonal migration patterns in sockeye salmon ( Oncorhynchus nerka ) and Atlantic salmon ( Salmo salar ). Atsumi and Koizumi [ 38 ] used X (formerly Twitter) and Google Images to explore spatial variations in breeding timing in Japanese dace fish ( Tribolodon hakonensis ) and how they may have been affected by climate change. Combined with data on breeding success or the costs of not adjusting breeding timing, these studies could greatly advance function-related research. Given the ongoing global environmental change, such explorations can be invaluable to understanding how these changes impact various species in terms of range shifts and/or expansions. Again, digital data has limits, and complementing it with traditional methods may be required to accurately assess the fitness value of a behavior.

The challenges and limitations of using digital data to study animal behavior

Addressing questions related to any of Tinbergen’s 4 levels of analysis is challenging. While digital data and approaches can greatly advance the fields of behavioral ecology and conservation behavior, these data sources and tools currently cannot replace empirical work and field studies. We acknowledge the limitations of digital data, particularly in answering questions related to internal mechanisms such as endocrine or neural control of behavior. Available digital data may not provide reliable information on an individual’s physiological state, its developmental history, or its reproductive state. Nonetheless, digital data sources can provide new opportunities to explore many aspects of Tinbergen’s 4 questions in a noninvasive and without manipulation of free-living animals, thus solving underlying ethical and welfare issues associated with the use of animals in research [ 39 ]. It is important to note, however, that digital data research also raises ethical questions and should follow rules to avoid disruption to the focal animal(s), the animals’ population, or the wider ecosystem. Viewing digital data as complementary to more traditional sources of data may be very useful. Moreover, in some areas traditional data sources are lacking, and so adequately reliable digital data may be the best source of behavioral data available. Nevertheless, we must consider the biases, technical challenges, and ethical concerns associated with digital data.

First, data sets obtained from online platforms—particularly ones provided by the general public—have an inherited bias linked to Internet coverage and use such that different regions of the world are not equally represented in digital records. Similarly, different sectors of society based on, for example, ethnicity, language, socioeconomic status, and education level, are currently not equally represented in the digital realm, complicating research on human–nature interactions using digital data.

Second, only a fraction of the global biodiversity is digitally recorded and has an online presence [ 1 , 40 ]. This limits the number of species that can be explored using digital data sets and leads to an uneven sampling effort across different taxa and clades. Such biases, for example, towards charismatic or larger-bodied species, are widespread and well known from more traditional approaches of scientific research [ 41 ], but may be exacerbated using data from social media. Furthermore, this limitation of unequal human interest goes beyond which species are predominantly documented, but also to which behaviors are recorded. Such human preferences and biases, and how they may differ across cultures, may compromise analyses and conclusions if not properly accounted for [ 42 ]. Furthermore, search algorithms of search engines like Google or platform-internal ones may also introduce biases affecting the results returned.

The lack of rigorous collection protocols across various digital platforms, especially in light of the complexity and variety of animal behavior, makes applying digital data sources in behavioral ecology research even more challenging. For example, in exploring bird plumage color aberrations using various digital sources (Google Images and several local platforms devoted to bird watching and photography), Zbyryt and colleagues [ 32 ] highlighted how digital sources and public participation can advance our understanding of less-studied natural phenomena. They showed that color aberrations are more prevalent in urban, larger, and sedentary birds. However, the nature of the input data prevented them from concluding whether these patterns were biologically driven or resulted from inherent biases in their data set that people more easily spot and report large sedentary birds in human settlements. Thus, it is essential to address these and other biases and limitations to understand when and where it is appropriate to use various digital data sources. As a start, combining data from novel digital sources—such as various social media platforms and Google Images—with more rigorous scientific data sets, dedicated fieldwork, or literature surveys, can help validate digital sources and ensure meaningful results. Another approach is creating well-designed question-first citizen science data sets in which researchers recruit and train citizen scientists to collect dedicated data to answer specific questions [ 43 ].

When exploring user-generated content—for example, videos uploaded on social media platforms—we must also consider legal and ethical aspects such as data protection and privacy [ 44 ]. In order to minimize the risk of misusing sensitive data (e.g., IP address, localization details, or user name), we advocate for establishing and following protocols for data protection [ 44 ]. It is also important to note that many social media recordings may be associated with unintentional or even intentional disturbances and harmful actions towards the animal being recorded [ 5 , 45 ], raising ethical concerns as well as questions of interpretation and relevance. Even if individuals are not directly harmed, the context under which data were recorded (e.g., Were domestic animals like dogs present? Did the humans feed the animals before filming?) is not always known, and this may have substantial impacts on the recorded behavior [ 17 ]. Such human disturbances, combined with partial recording and suboptimal recording quality, necessitate extensive filtering processes and the implementing of clear protocols for the inclusion of records. Furthermore, it may limit the use of digital data sources in certain explorations [ 7 ]. While we encourage people to share their nature observations online, we discourage harmful human–nature interactions to obtain these observations. By contrast, recording people’s negative interactions with nature can potentially be helpful for both legal and conservation interventions, as well as for related research.

Finally, while these readily available data sets are relatively easy to obtain, using them requires programming skills, computational power, and storage capacity, among other things [ 46 ]. Accessing various platforms may further require data-sharing agreements, proprietary companies opening their data sets for researchers, and consistency in how data is managed [ 47 ]. Once obtained, data filtering and cleaning processes and analysis would further require advanced technological tools, such as machine learning methods and machine vision models. Such filtering process should also consider for example AI-generated content and ensure only reliable data are used. Post-analysis challenges may include repeatability and reproducibility [ 4 , 48 ] as data may not be archived on different platforms, and downloading and sharing all records may face legal issues (copyrights), as well as storage space limitations. While some of these aspects are beyond our control, keeping clear records of protocols, versions, and codes, as well as publishing metadata and when possible raw data, could increase transparency and help address some limitations [ 4 ].

Conclusions and future outlook

The use of digital data in ecological and evolutionary research on animal behavior has emerged as a promising approach to enhance traditional data sources and overcome several constraints such as lack of time, accessibility, and financial resources. Digital data enables researchers to conduct retrospective analysis and comparisons across various temporal, spatial, and taxonomic scales, providing a potentially vast data set to explore. Moreover, as Internet use continues to grow and new digital platforms emerge, more data will become available, offering further opportunities to advance both basic and applied studies in behavioral ecology. The use of digital data in behavioral ecology is rapidly increasing and will potentially unveil larger data sets and larger audiences than existing citizen science platforms [ 35 , 49 ]. These new databases will enable researchers to ask basic and novel questions and study animal behavior with greater depth and scope. Furthermore, by leveraging social media data created by individuals, researchers can advance knowledge on animal (including human) behavior, promote public engagement with nature, and enhance present and future conservation efforts.

In addition to using data already uploaded to the Internet, scientists can encourage people to upload data containing species or areas of interest for their study. Researchers can also recruit people to help filter, score, or tag data collected online as on the Zooniverse platform ( https://www.zooniverse.org/ ), with the ultimate goal of involving the public in biodiversity conservation and science and facilitating the processing of big data. With advances in AI models, such collection and classification of data can be made automatically (fully or semi), based on taxonomic group or the location where the observation was recorded. This will enhance the ability of researchers to incorporate publicly available data in their studies. For example, using machine learning approaches, Pardo and Wittemyer [ 50 ] were able to find a name-like calling behavior in African savannah elephants ( Loxodonta africana ). However, limited by their sample size, they were not able to isolate and encode specific “name” sounds. Social media recording of tourists in those areas could potentially help in future research.

With the increasing global environmental challenges linked to biodiversity loss and climate change, digital resources are invaluable sources of data, especially in time-sensitive cases. Behavioral aspects such as interspecific interactions or behavioral flexibility are missing from many large-scale analyses and predictions of future species responses to human-driven environmental changes [ 51 , 52 ]. Digital data can greatly improve our ability to successfully integrate such behavioral dimensions into spatial modeling of abiotic changes and help us produce more realistic estimates of future risks and potential species distributions [ 52 ]. Taken together, such studies can help us develop a rich understanding of behavior based on the Tinbergen framework.

From an applied perspective, the field of conservation behavior [ 53 ] can benefit substantially from digital data sources too. Online images and video repositories can help conservation scientists and managers better understand anthropogenic impacts on animal behavior, identify behavioral indicators of changes to the species’ environment, highlight potential human–wildlife conflicts, and design and implement behavior-sensitive management [ 54 ]. With the great advancements in AI and machine learning and the increased availability of big data, we expect that more behavioral ecologists and conservation scientists will start incorporating digital-based data sources and approaches alongside their field and empirical work.

Supporting information

S1 table. examples of publications utilizing digital data for behavioral ecology divided into their potential contribution to understanding animal behavior according to tinbergen’s 4 questions..

https://doi.org/10.1371/journal.pbio.3002793.s001

• View Article
• PubMed/NCBI
• Google Scholar

• YouTube icon

Evolution News & Science Today

Neuroscience & mind, there is no known evolutionary rule for animal intelligence.

Cambridge neuropsychologist  Nicholas Humphrey argues that warm-bloodedness  (endothermy)  enables mammals and birds to be more sentient than, say, cold-blooded reptiles and fish. In an  excerpt  from his book,  Sentience:  The Invention of Consciousness  (MIT Press 2023), he argues that that development put these endotherms on the road to consciousness.

A Surprising Fact About Warm-Bloodedness

Humphrey points out that while warm-bloodedness is metabolically expensive — we need lots of food to maintain a high body temperature — it’s not a stepwise expense by any means:

For one thing, as temperature goes up various bodily processes actually become more energetically efficient, so the costs can be partially offset. In particular, the cost of sending an impulse along a nerve decreases until it reaches a minimum at about 37 degrees Celsius. The result is that, although the overall running costs for the body go up with being warm-blooded, the costs for the brain are reduced. This means that mammals and birds can support larger and more complex brains with relatively little extra outlay of energy. NICHOLAS HUMPHREY, “DID WARM-BLOODEDNESS PAVE THE  PATH TO SENTIENCE?, ” MIT PRESS, APRIL 15, 2024

He asks us to consider how warm-bloodedness might specifically affect the qualities we associate with intelligence:

It’s a well-established fact of physiology that the functional characteristics of neurons change with temperature. It’s been found for a range of animals — warm and cold-blooded — that the conduction speed for all classes of neurons increases by about 5 percent per degree Centigrade, while the refractory period decreases by roughly the same amount. This implies that when the ancestors of mammals and birds transitioned from a cold-blooded body temperature of, say, 15 degrees Celsius (59 degrees Fahrenheit) to a warm-blooded temperature of 37 degrees Celsius, the speed of their brain circuits would have more than doubled. We’ve remarked already on the “lucky accidents” that have, at several points, played a part in the evolution of sensations. If warm-bloodedness played these key roles, first in changing the way animals thought about the autonomy of the self, second in preparing the brain for phenomenal consciousness, here was an accident as lucky as they come. HUMPHREY, “ PATH TO SENTIENCE? ”

But the Situation Is Not Clear-Cut

When researchers have tested reptiles, using the same tests used on birds, the results have been surprising:

The lizards’ success on a worm-based test normally used on birds was “completely unexpected,” said Duke biologist Manuel Leal, who led the study. He tested the lizards using a wooden block with two wells, one that was empty and one that held a worm but was covered by a cap. Four lizards, two male and two female, passed the test by either biting the cap or bumping it out of the way. The lizards solved the problem in three fewer attempts than birds need to flip the correct cap and pass the test, Leal said. Birds usually get up to six chances a day, but lizards only get one chance per day because they eat less. In other words, if a lizard makes a mistake, it has to remember how to correct it until the next day, Leal said. He and Duke graduate student Brian Powell describe the experiment and results online in  Biology Letters. ASHLEY YEAGER, “BRAINY LIZARDS PASS  TESTS FOR BIRDS, ” DUKE TODAY, JULY 12, 2011

Significantly, the lizards had to  learn  a new task. They don’t feed themselves in the wild by flipping caps. And when the researchers raised the bar by switching which well held the worm, two of the lizards figured it out. As a result, the researchers named them Plato and Socrates.

That’s hardly the only instance of reptile intelligence observed by researchers. For example, the  New York Times  interviewed comparative psychologist  Gordon M. Burghardt  on monitor lizards:

Other studies have documented similar levels of flexibility and problem solving. Dr. Burghardt, for instance, presented monitor lizards with an utterly unfamiliar apparatus, a clear plastic tube with two hinged doors and several live mice inside. The lizards rapidly figured out how to rotate the tube and open the doors to capture the prey. “It really amazed us that they all solved the problem very quickly and then did much better the second time,” Dr. Burghardt said. “That’s a sign of real learning.”  EMILY ANTHES, “COLDBLOODED  DOES NOT MEAN  STUPID,” NEW YORK TIMES, NOVEMBER 19, 2013

Training a Reptile

Anoles and monitor lizards are considered to be among the  most intelligent lizards . But it hasn’t been customary until recently to credit  any  lizards with much intelligence. That may be due to mishandling in some cases. Anthes notes at the  Times  that tests for reptile intelligence should take into account  normal differences  between, say, mammal behavior and reptile behavior: “By using experiments originally designed for mammals, researchers may have been setting reptiles up for failure. For instance, scientists commonly use “aversive stimuli,” such as loud sounds and bright lights, to shape rodent behavior. But reptiles respond to many of these stimuli by freezing, thereby not performing.”

Warm-bloodedness enables the mammal and the bird to be active for much longer than the reptile and to be active in colder temperatures. Thus, mammals and birds likely encounter more situations where they can (and must) demonstrate sentience and intelligence. But warm-bloodedness does not seem to be an essential component of those qualities.

Sorry, there was a problem.

Download the free Kindle app and start reading Kindle books instantly on your smartphone, tablet, or computer - no Kindle device required .

Read instantly on your browser with Kindle for Web.

Using your mobile phone camera - scan the code below and download the Kindle app.

Image Unavailable

• To view this video download Flash Player

Follow the author

Animal Intelligence: Experimental Studies (Classic Reprint) Paperback – August 24, 2018

• Print length 318 pages
• Language English
• Publisher Forgotten Books
• Publication date August 24, 2018
• Dimensions 9 x 6 x 0.39 inches
• ISBN-10 1440070423
• ISBN-13 978-1440070426
• See all details

Product details

• Publisher ‏ : ‎ Forgotten Books (August 24, 2018)
• Language ‏ : ‎ English
• Paperback ‏ : ‎ 318 pages
• ISBN-10 ‏ : ‎ 1440070423
• ISBN-13 ‏ : ‎ 978-1440070426
• Item Weight ‏ : ‎ 15.2 ounces
• Dimensions ‏ : ‎ 9 x 6 x 0.39 inches
• #7,804 in Medical General Psychology
• #40,830 in Psychology & Counseling

About the author

Edward l. thorndike.

Discover more of the author’s books, see similar authors, read author blogs and more

Customer reviews

• 5 star 4 star 3 star 2 star 1 star 5 star 100% 0% 0% 0% 0% 100%
• 5 star 4 star 3 star 2 star 1 star 4 star 100% 0% 0% 0% 0% 0%
• 5 star 4 star 3 star 2 star 1 star 3 star 100% 0% 0% 0% 0% 0%
• 5 star 4 star 3 star 2 star 1 star 2 star 100% 0% 0% 0% 0% 0%
• 5 star 4 star 3 star 2 star 1 star 1 star 100% 0% 0% 0% 0% 0%

Customer Reviews, including Product Star Ratings help customers to learn more about the product and decide whether it is the right product for them.

To calculate the overall star rating and percentage breakdown by star, we don’t use a simple average. Instead, our system considers things like how recent a review is and if the reviewer bought the item on Amazon. It also analyzed reviews to verify trustworthiness.

• Sort reviews by Top reviews Most recent Top reviews

Top reviews from the United States

There was a problem filtering reviews right now. please try again later..

Top reviews from other countries

• About Amazon
• Investor Relations
• Amazon Devices
• Amazon Science
• Sell products on Amazon
• Sell on Amazon Business
• Sell apps on Amazon
• Become an Affiliate
• Advertise Your Products
• Self-Publish with Us
• Host an Amazon Hub
• › See More Make Money with Us
• Amazon Business Card
• Shop with Points
• Reload Your Balance
• Amazon Currency Converter
• Amazon and COVID-19
• Your Account
• Your Orders
• Shipping Rates & Policies
• Returns & Replacements
• Manage Your Content and Devices

• Conditions of Use
• Privacy Notice
• Consumer Health Data Privacy Disclosure
• Your Ads Privacy Choices

• Skip to main content
• Skip to "About this site"

Language selection

• Français
• Search and menus

Analytical Studies Branch Research Paper Series Experimental Estimates of Potential Artificial Intelligence Occupational Exposure in Canada

DOI : https://doi.org/10.25318/11f0019m2024005-eng

Skip to text

Acknowledgements

Executive summary, 1 introduction, 4 conclusion.

Text begins

The authors would like to thank Li Xue, Marc Frenette and Vincent Hardy from Statistics Canada, and Jessica Gallant, Matthew Calver, Jacob Loree and Alan Stark from the Department of Finance Canada for their helpful and constructive comments.

Past studies on technological change have suggested that occupations involving routine and manual tasks will face a higher risk of automation-related job transformation. However, recent advances in artificial intelligence (AI) challenge prior conclusions, as AI is increasingly able to perform non-routine and cognitive tasks. These advances have the potential to affect a broader segment of the labour force than previously thought. This study provides experimental estimates of the number and percentage of workers in Canada potentially susceptible to AI -related job transformation based on the complementarity-adjusted AI occupational exposure index of Pizzinelli et al. (2023), inspired by Felten, Raj and Seamans (2021). Results from the 2016 and 2021 censuses of population suggest that, on average, about 60% of employees in Canada could be exposed to AI -related job transformation, and about half of this group are in jobs that may be highly complementary with AI . Unlike previous waves of automation, which mainly transformed the jobs of less educated employees, AI is more likely to transform the jobs of highly educated employees. Despite facing potentially higher exposure to AI -related job transformation, highly educated employees may be in jobs that could benefit from AI technologies. Compared with employees in other industries, exposure to AI -related job transformation is higher for employees in professional, scientific and technical services; finance and insurance; information and cultural industries; educational services; and health care and social assistance. However, education and health care professionals are more likely to be in jobs that are highly complementary with AI . Employees in industries such as construction, and accommodation and food services face relatively lower exposure to AI -related job transformation. Whether occupations that may benefit from AI will experience relatively higher employment and wage growth remains to be seen, as this depends on factors such as firm productivity and the ability of workers in those occupations to leverage the potential benefits of AI .

Recent developments in the field of artificial intelligence (AI) have fuelled excitement, as well as concerns, regarding its implications for society and the economy. While previous waves of technological transformation raised concerns regarding the future of jobs involving routine and manual tasks, a broader segment of the labour force could be affected in an era when sophisticated large language models such as ChatGPT increasingly excel at performing non-routine and cognitive tasks typically done by highly skilled workers. AI encompasses a lot more than just natural language processing. These technologies not only have the capacity to automate routine tasks but can also augment human decision-making processes and create entirely new opportunities for innovation and efficiency. As AI continues to evolve, it has the potential to reshape industries, redefine job roles and transform the nature of work. With the transformative effects of AI already in motion, it raises renewed concerns about job transformation and the need for workforce adaptation.

This study adopts the complementarity-adjusted AI occupational exposure index of Pizzinelli et al. (2023), inspired by the original AI occupational exposure measure of Felten, Raj and Seamans (2021), and applies it to data from the 2016 and 2021 censuses of population. The experimental estimates presented in this study are largely based on the technological feasibility of automating job tasks. Employers may not immediately replace human labour with AI , even if it is technologically feasible to do so, because of financial, legal and institutional constraints. Consequently, exposure to AI does not necessarily imply a risk of job loss. At the very least, it could imply a certain degree of job transformation (Frenette and Frank, 2020). Additionally, some economists argue that the risks and benefits currently being attributed to AI may be exaggerated (Acemoglu and Johnson, 2024; McElheran et al. , 2024), and productivity increases at the macroeconomic level may be modest at best (Acemoglu, 2024).

Following Pizzinelli et al. (2023), this study groups occupations into three categories based on their exposure to and complementarity with AI : (1) high exposure and low complementarity, (2) high exposure and high complementarity, and (3) low exposure. Results suggest that in May 2021, on average, around 4.2 million employees (31%) in Canada were in the first group, about 3.9 million (29%) were in the second group and about 5.4 million (40%) were in the third group. This distribution was very similar in May 2016. Unlike previous waves of automation, which mainly transformed the jobs of less educated employees performing routine and non-cognitive tasks, AI is more likely to transform the jobs of highly educated employees performing non-routine and cognitive tasks. However, highly educated employees are also more likely to hold jobs that are highly complementary with AI technologies than less educated employees. But workers will still need the skills to be able to leverage the potential benefits of AI . Compared with employees in other industries, exposure to AI -related job transformation is higher for employees in professional, scientific and technical services; finance and insurance; information and cultural industries; educational services; and health care and social assistance. However, education and health care professionals are more likely to be in jobs highly complementary with AI . Employees in industries such as construction, and accommodation and food services face relatively lower exposure to AI -related job transformation.

There is a lot of uncertainty when it comes to predicting the transformative effects of technological changes on the labour market. This study provides a static picture of AI occupational exposure based on employment compositions in May 2016 and May 2021, which were fairly similar. How workers respond and adapt to the potentially evolving labour market in the long run remains to be seen. The index used in this study is subjective and based on judgments regarding some current possibilities of AI . Consequently, the applicability of the index may decrease over time as AI capabilities grow and AI can perform an increasing number of tasks currently done by human workers. Alternative measures of AI exposure could provide further insights. Future research could also attempt to answer the question, “What happened to workers whose jobs were exposed to AI -related job transformation?”

A couple of centuries ago, the Industrial Revolution and the forces of globalization coalesced to fundamentally change the global economy. These forces served as catalysts for the technological progress that has been a cornerstone of economic development. Technological advancements and innovation paved the way for machines to take over some labour-intensive tasks and allowed workers to focus on more cognitive tasks requiring creativity and critical thinking. Adoption of new technologies also led to the obsolescence of some jobs, serving as a pathway toward higher productivity. A prominent example of this is the advent of computers, which undoubtedly replaced some jobs but also created new ones in the process (see, e.g. , Autor, Levy and Murnane [2003] or Graetz and Michaels [2018]). However, higher productivity may not always translate to higher wages for workers (Acemoglu and Johnson, 2024).

More generally, automation has become a defining feature of modern economies, including Canada’s. It has revolutionized various industries by streamlining processes, increasing efficiency and reducing operational costs, among other things. It has also raised concerns about the future of workers. The widely cited study by Frey and Osborne (2013), which estimated automation risks in the United States, has spurred a growing body of literature surrounding automation (see, e.g. , Arntz, Gregory and Zierahn [2016]; Oschinski and Wyonch [2017]; Nedelkoska and Quintini [2018]; Frenette and Frank [2020]; and Georgieff and Milanez [2021]). Frenette and Frank (2020) estimated that approximately 1/10 of employees in Canada could be at high risk (probability of 70% or higher) of automation-related job transformation.

The prevailing thought from the automation literature is that highly educated or highly skilled individuals are less susceptible to automation-related job transformation because they are more likely to perform non-routine and cognitive tasks, which are thought to be less automatable. However, another source of disruption, which has the potential to upend prior notions, is emerging: artificial intelligence (AI) . Note While AI has been around for decades ( e.g. , video games, image recognition), it was not until 2022 when it became mainstream and surged in popularity, partly fuelled by the release of ChatGPT by OpenAI.

The unprecedented pace of advancements in the field of AI and its increasing integration into society and the economy have led some researchers to call this a pivotal moment in history, akin to the transformative shifts brought on by the Industrial Revolution (Cazzaniga et al. , 2024). ChatGPT is just one example of a large language model (LLM) that has unlocked the remarkable possibilities of AI . AI can also perform complex tasks like generating music and videos from text input ( e.g. , Sora by OpenAI). AI encompasses a wide range of applications, including natural language processing, machine learning, computer vision and robotics. These technologies not only have the capacity to automate routine tasks but can also augment human decision-making processes and create entirely new opportunities for innovation and efficiency. As the field of AI continues to evolve, it has the potential to reshape industries, redefine job roles and transform the nature of work. In today’s rapidly evolving technological landscape, the integration of AI into various aspects of society, from virtual assistants and recommendation algorithms to autonomous vehicles and predictive analytics, questions naturally arise regarding its impact on society and the economy. The widespread adoption of AI raises renewed concerns about job transformation, skill mismatches and the need for workforce adaptation.

The primary objective of this study is to quantify the level of potential AI occupational exposure (AIOE) in Canada. By employing experimental methods, this study offers preliminary insights into how AI may affect the Canadian labour market and the potential risks and benefits it holds for workers.

This study adopts the complementarity-adjusted AIOE (C-AIOE) index proposed by Pizzinelli et al. (2023). The original AIOE index, which is often cited in the literature, was proposed by Felten, Raj and Seamans (2021) as a way of measuring how AI applications overlap with the human abilities needed to perform a given job. In light of recent advancements in LLMs, Felten, Raj and Seamans (2023) considered an alternate index that weighted language modelling more heavily and found that it was highly correlated with the original AIOE index. Recognizing that AI can complement human labour, the International Monetary Fund (IMF) study by Pizzinelli et al. (2023) proposed the C-AIOE index, which attempts to account for the potential complementarity of AI across occupations, in addition to direct exposure. These measures focus on “narrow” AI , which refers to “computer software that relies on highly sophisticated algorithmic techniques to find patterns in data and make predictions about the future” (Broussard, 2018; Felten, Raj and Seamans, 2021). This definition encompasses generative AI ( e.g. , LLMs, image recognition) but does not capture exposure to “general” AI , which refers to “computer software that can think and act autonomously and is combined with automation and robot technologies” (Pizzinelli et al. , 2023). International comparisons of AIOE based on the original AIOE index have been done (see, e.g. , Georgieff and Hyee [2021] and OECD [2023]). An IMF study by Cazzaniga et al. (2024) compared AI exposure and potential complementarity across countries using the C-AIOE index but did not analyze Canadian data in detail. They found that around 60% of jobs in advanced economies may be highly exposed to AI -related job transformation. As will be shown, this is similar to the share estimated for Canada.

This study offers Canadian evidence on AIOE and asks the following research questions:

• Which occupations are potentially exposed to AI -related job transformation?
• Which occupations may benefit from AI -related job transformation?
• How does the distribution of AIOE vary by industry, education level, employment income and other worker characteristics?

The experimental AI exposure estimates in this study are largely based on the technological feasibility of automating job tasks. Employers may not immediately replace humans with AI , even if it is technologically feasible, for several reasons (see, e.g. , Bryan, Sood and Johnston [2024]), including financial, legal and institutional factors. Consequently, exposure to AI does not necessarily imply a risk of job loss. At the very least, it could imply some degree of job transformation (Frenette and Frank, 2020). AI could lead to the creation of new tasks within existing jobs or create entirely new jobs. Additionally, some economists argue that the risks and benefits of AI may be exaggerated (Acemoglu and Johnson, 2024; McElheran et al. , 2024), and productivity increases at the macroeconomic level may be modest at best (Acemoglu, 2024). Evidence from the United States suggests that the adoption of AI has been more prevalent in larger firms (McElheran et al. , 2024), as some employers may not yet find it economically optimal to adopt such technologies (Svanberg et al. , 2024). Whether this will contribute to a productivity gap between smaller and larger firms is unclear. Predicting the effects of technological changes on the labour market is not an exact science, as some subjectivity is usually involved. For example, more than a decade after Frey and Osborne (2013), it is still difficult to precisely measure the effect of automation on labour markets, as changes are ongoing (Georgieff and Milanez, 2021). Although the diffusion of new technology can take time (Feigenbaum and Gross, 2023), measuring the impact of AI could be challenging given the rapid pace of advancements. The experimental estimates presented in this study should be interpreted with caution. Only time will tell whether predicted changes brought on by new technologies will come to fruition.

The remainder of this article is organized as follows. Section 2 briefly describes the AIOE index of Felten, Raj and Seamans (2021) and the complementarity-adjusted variant of Pizzinelli et al. (2023). Section 3 presents the results, and Section 4 provides concluding remarks and suggestions for future research.

The objective of this study is to estimate the extent to which occupations in Canada are potentially exposed to AI -related job transformation and the extent to which AI can potentially complement human labour in those occupations. This study uses the novel C-AIOE index of Pizzinelli et al. (2023) to achieve this objective. This measure is computed at the occupational level based on data from the Occupational Information Network (O*NET), which was created in the late 1990s by the United States Department of Labor to quantify and track the skills and abilities used across more than 1,000 different occupations ( https://www.onetonline.org ). Thus, the measure used in this study relies on occupational attribute data from the United States, which has a similar skill profile as Canada.

The C- AIOE index is based on the original AIOE index of Felten, Raj and Seamans (2021), which measures the relationship between 52 human abilities and 10 AI applications, weighted by the degree of complexity and importance of those skills for a given occupation i MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGPbaaaa@3704@ ,

where j MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGQbaaaa@3705@ indexes 52 occupational abilities; L j i MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGmbWdamaaBaaaleaapeGaamOAaiaadMgaa8aabeaaaaa@391E@ is the prevalence score from O*NET and I j i MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGjbWdamaaBaaaleaapeGaamOAaiaadMgaa8aabeaaaaa@391B@ is the importance score from O*NET for ability j MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGQbaaaa@3705@ in occupation i MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGPbaaaa@3704@ ; and A j = ∑ k = 1 10 x k j MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGbbWdamaaBaaaleaapeGaamOAaaWdaeqaaOWdbiabg2da9maa wahabeWcpaqaa8qacaWGRbGaeyypa0JaaGymaaWdaeaapeGaaGymai aaicdaa0WdaeaapeGaeyyeIuoaaOGaamiEa8aadaWgaaWcbaWdbiaa dUgacaWGQbaapaqabaaaaa@4353@ is the exposure to AI of ability j MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGQbaaaa@3705@ computed as the sum of the relatedness scores, x k j MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWG4bWdamaaBaaaleaapeGaam4AaiaadQgaa8aabeaaaaa@394C@ , of ability j MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGQbaaaa@3705@ with 10 AI applications. Note This index is a relative measure of AI exposure ( e.g. , A I O E m >   A I O E n MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbiqaaWVbqaaaaa aaaaWdbiaadgeacaWGjbGaam4taiaadweapaWaaSbaaSqaa8qacaWG TbaapaqabaGcpeGaeyOpa4JaaeiiaiaadgeacaWGjbGaam4taiaadw eapaWaaSbaaSqaa8qacaWGUbaapaqabaaaaa@41DD@ implies that occupation m MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGTbaaaa@3708@ faces greater exposure to AI -related job transformation than occupation n MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGUbaaaa@3709@ ). See Felten, Raj and Seamans (2021) for details.

Because the AIOE index is agnostic regarding the implications of occupations being exposed to AI , Pizzinelli et al. (2023) proposed a variant of the AIOE index that accounts for the potential complementarity of AI . They make the case that certain occupations may be less conducive to the unsupervised use of AI than others. For example, judges and medical professionals are examples of occupations where job aspects such as the criticality of decisions and the gravity of the consequences of errors may require human workers to make the final decision (Cazzaniga et al. , 2024). The C-AIOE of Pizzinelli et al. (2023) is computed as

where 0 ≤ w ≤ 1 MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaaIWaGaeyizImQaam4DaiabgsMiJkaaigdaaaa@3BF1@ is a weight chosen by the researcher that controls the influence of the complementary parameter ( θ MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacqaH4oqCaaa@37CC@ ), θ i MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacqaH4oqCpaWaaSbaaSqaa8qacaWGPbaapaqabaaaaa@3914@ is the complementarity index of occupation and i MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGPbaaaa@3704@ and θ M I N MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacqaH4oqCpaWaaSbaaSqaa8qacaWGnbGaamysaiaad6eaa8aabeaa aaa@3A99@ is the minimum observed θ MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacqaH4oqCaaa@37CC@ value among all occupations. A weight of w =   0 MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWG3bGaeyypa0Jaaeiiaiaaicdaaaa@3975@ reverts the C-AIOE back to the original AIOE ( e.g. , no role for AI complementarity), while w =   1 MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWG3bGaeyypa0Jaaeiiaiaaigdaaaa@3976@ allows maximum potential AI complementarity for occupation i MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGPbaaaa@3704@ . Note Like the AIOE index, the complementarity index is also a relative measure, with a higher value indicating higher potential complementarity. The complementarity index for occupation i MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacaWGPbaaaa@3704@ , θ i MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaaeaaaaaaaaa8 qacqaH4oqCpaWaaSbaaSqaa8qacaWGPbaapaqabaaaaa@3914@ , is computed using O*NET data on “work contexts” and “job zones” for that particular occupation. To do so, 11 work contexts (each score ranging from 0 to 100) and the job zone (ranging from 1 to 5) are combined into six components as follows:

• Face to face
• Public speaking

Although AI can play a role in enhancing certain aspects of communication, the nuanced complexities of face-to-face interactions and public speaking could remain predominantly within the realm of human expertise.

• For outcomes
• For others’ health

AI has the potential to transform many sectors in the economy, including health care, where tough decisions are routinely made, and such decisions may still require human oversight and judgment.

• Exposure to outdoor environments
• Physical proximity to others

Jobs requiring substantial outdoor exposure and proximity to others require a certain level of adaptability and teamwork ( e.g. , firefighters, construction workers). Integrating AI into highly advanced machinery in diverse work environments could be costly.

• Consequence of errors
• Freedom of decisions
• Frequency of decisions

The importance of human oversight may become increasingly evident as AI continues to automate decision-making processes. In professions such as air traffic control or nursing, where human judgment is paramount, the combination of data analysis and instinct is essential for responding to unexpected scenarios. While AI can offer valuable data and recommendations, thereby potentially reducing human error and accelerating decision making, the indispensability of human oversight remains clear.

• Degree of automation (100 minus the O*NET score so that occupations with a low degree of automation receive higher values)
• Unstructured versus structured work

Occupations involving routine tasks have historically been more susceptible to technological transformation. Despite differences between AI and previous waves of automation, routine-intensive occupations remain particularly vulnerable to transformation. In contrast, less structured jobs may necessitate more advanced technologies for AI to operate autonomously.

Job zone is an indicator of the extent of preparation required for a job. This value must be rescaled to align with the five other components by multiplying it by 20, so that it ranges from 20 to 100 instead of 1 to 5. A higher value indicates more extensive preparation.

Occupations with high educational or training requirements may be more conducive to integrating the skills complementary with AI , as providing instructions to AI and leveraging it require some level of expertise and proficiency.

A score for each of the six components is computed by averaging the work contexts within each component ( e.g. , the score for communication is the average of face-to-face and public speaking work contexts). For the skills component, the score is the rescaled job zone value. Then, θ is calculated as the average of the six component scores divided by 100. See Pizzinelli et al. (2023) for more details regarding the derivation of the C-AIOE index and the sensitivity analyses.

This index does have some limitations, as pointed out by Pizzinelli et al. (2023). The selection of O*NET variables that serve as inputs of the index is subjective and relies on judgment regarding the factors that matter for the interaction between AI and human workers. However, Pizzinelli et al. (2023) show that the work contexts are not all systematically related to each other and offer a multifaceted take on the potential complementarity of AI with human workers. The index considers how human abilities may overlap with 10 AI applications, but as AI capabilities improve, the likelihood of AI supplanting tasks typically performed by human workers may grow. Consequently, the applicability of the index could decrease over time. Note Moreover, while the index captures the potential exposure of occupational abilities and tasks to AI , it does not account for advances in robotics, sensors and other technologies that could potentially integrate with AI (Felten, Raj and Seamans, 2021).

As O*NET is an American database, the occupations are coded according to the Standard Occupational Classification (SOC) system. The complementarity parameter and the AIOE index were computed based on version 28.2 of the O*NET database, which uses the 2018 SOC . The AIOE index was computed at the six-digit level, while the complementarity parameter was computed at the eight-digit level and then aggregated to the six-digit level by averaging the parameter values ( e.g. , the values associated with SOC codes 12-3456.01 and 12-3456.02 would be averaged to obtain the value for SOC code 12-3456). The six-digit SOC codes were then converted to the four-digit codes of version 1.3 of the Canadian National Occupational Classification (NOC) 2016 so the rich set of dimensions from the 2016 and 2021 censuses of population (reference week in May) could be used to examine AIOE in Canada. Note The sample was restricted to employees aged 18 to 64 living off reserve in private dwellings, excluding full-time members of the Canadian Armed Forces. Employment in some industries, such as accommodation and food services, decreased from May 2016 to May 2021 because of the COVID-19 pandemic, so the 2016 Census of Population was also used as a robustness check. However, results suggest that the share of employees exposed to AI -related job transformation changed very little in general.

Figure 1 presents the AIOE and potential complementarity ( θ ) for Canadian occupations. The median AIOE was around 6.0, while the median complementarity was about 0.6. Following Pizzinelli et al. (2023), an occupation is considered “high exposure” if its AIOE exceeds the median AIOE and “low exposure” otherwise. Likewise, an occupation is considered “high complementarity” if its potential complementarity exceeds the median complementarity and “low complementarity” otherwise. Note Based on this, occupations are grouped into four quadrants in Figure 1: high exposure and low complementarity, high exposure and high complementarity, low exposure and low complementarity, and low exposure and high complementarity. For simplicity, the latter two categories are combined into a single category, “low exposure,” in subsequent analyses. High-exposure, low-complementarity occupations are those that may be highly exposed to AI -related job transformation and whose tasks could be replaceable by AI in the future. High-exposure, high-complementarity occupations are those that may be highly exposed to AI -related job transformation but could be highly complementary with AI . However, workers will still need the necessary skills to leverage the complementary benefits of AI . Low-exposure jobs are those that may be less exposed to AI -related job transformation than others. Note

Potential artificial intelligence occupational exposure (AIOE) and complementarity in Canada

This chart shows a scatter plot with the AI occupational exposure index ranging from 5 to 7 on the horizontal axis and the complementarity index ranging from 0.4 to 0.8 on the vertical axis. There are 490 data points. Each data point represents an occupation as per the 4-digit National Occupation Classification version 2016 and are colour-coded with three different colours. The colours are used to distinguish the occupations according to their minimum educational requirement. Occupations requiring a bachelor's degree or higher are represented by blue, occupations requiring some postsecondary education below bachelor's degree are represented by green, and occupations requiring high school or less education are represented by red. The chart shows the relationship between AI occupational exposure and the extent to which AI can play a complementary role in a given occupation. A higher AI occupational exposure index is associated with greater potential occupational exposure to AI . A higher complementarity index is associated with greater potential complementarity with AI . The median AI occupational exposure index score of 6 and the median complementarity index score of 0.6 are used to group the various occupations into four quadrants. The top-left quadrant contain data points representing occupations which might be relatively less exposed to AI and highly complementary with AI . The majority of occupations in that quadrant require some postsecondary education below bachelor's degree but there are also a few which require high school or less education. Some examples include firefighters, plumbers, and carpenters. The bottom-left quadrant contain data points representing occupations which might also be relatively less exposed to AI but also less complementary with AI . The majority of occupations in that quadrant require high school or less education but there are also a few which require some postsecondary education below bachelor's degree. Some examples include food and beverage servers, labourers in processing, manufacturing and utilities, and welders and related machine operators. The top-right quadrant contain data points representing occupations which might be highly exposed to AI and highly complementary with AI . The majority of occupations in that quadrant require a bachelor's degree or higher education but there are a few which require some postsecondary education below bachelor's degree. Some examples include general practitioners and family physicians, secondary school teachers, and electrical engineers. The bottom-right quadrant contain data points representing occupations which might be highly exposed to AI but less complementary with AI . This quadrant has fewer data points than the other quadrants and the occupations represented by the data points have a mixture of educational requirements. Some examples include data entry clerks, economists, computer network technicians, and computer programmers and interactive media developers.

Notes: The AIOE index and potential complementarity are based on Felten, Raj and Seamans (2021) and Pizzinelli et al. (2023). An occupation is considered high-exposure if its AIOE index exceeds the median AIOE across all occupations (6.0) and considered low-exposure otherwise. Similarly, an occupation is considered high-complementarity if its complementarity parameter exceeds the median complementarity across all occupations (0.6) and considered low-complementarity otherwise. Occupations in this chart are based on the 4-digit National Occupational Classification (NOC) 2016 version 1.3 converted from the United States Standard Occupational Classification (SOC) 2018. Of the 500 NOC occupations, 10 occupations which represented less than 1% of Canadian employment, were excluded due to a lack of Occupational Information Network (O*NET) data for computing the AIOE or complementarity parameter.

Source: Occupational Information Network (O*NET) version 28.2.

Figure 1 shows that jobs potentially highly exposed to AI -related job transformation are generally those that require higher education. Although these jobs could face relatively more exposure to AI -related transformation, occupations such as family physicians, teachers and electrical engineers may be complementary with AI technologies given their relatively high complementarity scores. In contrast, occupations such as computer programming, which may also require relatively high education, have low complementarity scores, suggesting less potential complementarity with AI . There is considerable uncertainty, however, regarding the extent to which AI can actually replace human labour.

Low-exposure occupations appear to be those that usually do not require a high level of education. Some examples of occupations facing relatively low exposure to AI -related job transformation are carpenters; welders; plumbers; food and beverage servers; labourers in processing, manufacturing and utilities; and firefighters. However, as illustrated by Figure 1, AI has the potential to transform a broad set of occupations regardless of skill level. The diffusion of AI could also have downstream general equilibrium effects. For example, although less educated employees may be in jobs potentially less exposed to AI -related job transformation, highly educated employees from high-exposure jobs could transition to low-exposure jobs, displacing less educated employees (see, e.g. , Beaudry, Green and Sand [2016]).

Chart 1 aggregates the various NOC occupations into 28 distinct jobs to simplify the analysis and precisely identify the number and distribution of employees falling into the three AI exposure groups: (1) high exposure and low complementarity, (2) high exposure and high complementarity, and (3) low exposure . In May 2021, on average, roughly 4.2 million employees (31%) in Canada were in the first group, about 3.9 million (29%) were in the second group and about 5.4 million (40%) were in the third group.

Chart 1 start

Chart 1 end

At least three-quarters of employees in the following occupations were in the first group ( i.e. , highly exposed to AI -related job transformation and whose tasks could be replaceable with AI in the future): administrative occupations in finance, insurance and business; office support and co-ordination occupations; sales representatives and salespersons in wholesale and retail trade; service representatives and other customer and personal services occupations; professional occupations in business and finance; and computer and information systems professionals. Interestingly, among the 28 occupations, computer and information systems professionals experienced the highest growth (39%) from May 2016 to May 2021. However, this does not necessarily mean that computer and information systems professionals will be in less demand in the future because of AI . While these professionals may be in high-exposure, low-complementarity jobs, they are integral to maintaining and improving the underlying AI infrastructure, and this may lead to the creation of new tasks or jobs. Around 85% of employees or more in management occupations, professional occupations in education services and professional occupations in health (except nursing), as well as engineers, were in the second group ( i.e. , potentially highly exposed to AI -related job transformation, but AI can complement human labour as long as the worker possesses the necessary skills). Some occupations that could be less susceptible to AI -related job transformation (third group) were support occupations in sales and service; trades helpers, construction labourers and related occupations; assisting occupations in support of health services; and natural resources, agriculture and related production occupations.

Chart 2 shows the AI exposure distribution by industry based on the North American Industry Classification System 2017, at the two-digit level. More than half of employees in the following industries were in high-exposure, low-complementarity jobs: professional, scientific and technical services; finance and insurance; and information and cultural industries. In contrast, educational services, and health care and social assistance employed proportionately more employees who may be beneficiaries of AI . Within the health care and social assistance industry, it is mostly the professional occupations ( e.g. , nurses, physicians) that may be complementary with AI technologies (Figure 1). Employees in industries such as accommodation and food services, manufacturing, construction, and transportation and warehousing may face relatively lower exposure to AI -related job transformation.

Chart 2 start

Chart 2 end

Employees in larger enterprises (in the commercial sector) may face relatively higher exposure to AI -related job transformation (Chart 3), compared with their counterparts in smaller enterprises. Roughly over one-third of workers in enterprises with 500 or more employees were in high-exposure, low-complementarity jobs in May 2016. This compares with 25% to 28% of workers in smaller enterprises. However, employees in larger enterprises were somewhat more likely to be in jobs complementary with AI than their counterparts in smaller enterprises.

Chart 3 start

Chart 3 end

Educational attainment has historically been one of the most important indicators of whether a worker will be resilient to technological shocks. The growing consensus from the labour economics literature is that less educated workers face a higher risk of automation-related job transformation than highly educated workers because the former group is more likely to perform routine and manual tasks that are more susceptible to being automated. However, Chart 4 shows that AI could affect a broader segment of the labour force than previously thought because it has the capacity to perform non-routine and cognitive tasks. Highly educated employees may face higher exposure to AI -related job transformation, as was shown in Figure 1. The highest shares of high-exposure, low-complementarity jobs are held by employees with a bachelor’s degree (37%) or a college, CEGEP or other certificate or diploma below a bachelor’s degree (36%), followed by those with a graduate degree (32%), high school or less education (25%), and an apprenticeship or trades certificate or diploma (15%). However, employees with a bachelor’s degree or higher were more likely to hold jobs that may be highly complementary with AI than those with an education below the bachelor’s degree level, as long as the potential beneficiaries of AI possess the necessary skills. Employees with an apprenticeship or trades certificate or diploma may be less exposed to AI -related job transformation, as 73% were in low-exposure occupations. However, as previously mentioned, a more nuanced view is that while less educated workers may face potentially lower exposure to AI -related job transformation, highly educated workers from high-exposure jobs may transition to low-exposure jobs, displacing less educated workers (see, e.g. , Beaudry, Green and Sand [2016]).

Chart 4 start

Chart 4 end

Many of the results presented so far are contrary to the findings on automation documented in the labour economics literature over the past two decades, raising concerns about the nexus of automation and AI . Frenette and Frank (2020) estimated that around 1/10 of employees in Canada were at high risk (probability of 70% or more) of automation-related job transformation in 2016. Chart 5 suggests that exposure to AI -related job transformation decreases as the risk of automation-related job transformation increases. The majority of employees (60%) in jobs at high risk of automation-related transformation were in jobs that may be least exposed to AI -related transformation (Chart 5). In contrast, 18% of employees in jobs at low risk (probability of less than 50%) of automation were in low-exposure jobs. However, although potentially highly exposed to AI -related job transformation, employees at a lower risk of automation-related job transformation hold jobs that could be highly complementary with AI . Jobs facing a moderate risk (probability of 50% to less than 70%) of automation-related transformation were most likely to be high-exposure, low-complementarity jobs. These findings are important, as they suggest that the distinction between manual and cognitive tasks and between repetitive and non-repetitive tasks used in the last two decades in labour economics to understand automation-related technological transformation may not apply to AI .

Chart 5 start

Chart 5 end

Like previous waves of technological transformation, AI has the potential to boost productivity. But this process can also exacerbate earnings inequality. Chart 6 shows the AI exposure distribution across employment income deciles. More than half of the jobs in the bottom half of the distribution were low-exposure jobs, while around 30% were high-exposure, low-complementarity jobs. The middle of the distribution may be the most vulnerable to AI -related job transformation, with around one-third of jobs being high exposure and low complementarity. Exposure to AI -related job transformation increases with employment income, but higher earners hold jobs that may be highly complementary with AI . Although the top decile had the highest share of jobs potentially exposed to AI -related job transformation, they also had the highest share of jobs (55%) that are highly complementary with AI . If higher earners can take advantage of the complementary benefits of AI , their productivity and earnings growth may outpace those of lower earners, and this could exacerbate earnings inequality (Cazzaniga et al. , 2024). However, the diffusion of AI could also potentially reduce earnings inequality if AI happens to adversely affect high-skill occupations (see, e.g. , Webb [2020]).

Chart 6 start

Chart 6 end

Canada’s record population growth, recently driven by international migration, raises questions about the future of jobs done by immigrants and non-permanent residents. In May 2016, recent immigrants (those who landed from 2011 to 2016) (29%) were just as likely as Canadian-born individuals (29%) to be in high-exposure, low-complementarity jobs (Chart 7). However, by May 2021, while the share of Canadian-born individuals in such jobs remained the same, the share of recent immigrants (those who landed from 2016 to 2021) in these jobs increased to 37%. This was partly driven by the fact that nearly 1/10 of permanent residents who landed from 2016 to 2021 were employed in computer and information systems professions in May 2021—occupations more likely to be high exposure and low complementarity. Less than 5% of permanent residents who landed from 2011 to 2016 were employed in these professions in May 2016. This increasing concentration of recent immigrants in computer and information systems professions has been documented by Picot and Mehdi (forthcoming). Another reason could be the (temporarily) falling share of employment in occupations adversely affected by the COVID-19 pandemic. Non-permanent residents were more likely to be in high-exposure, low-complementarity jobs and low-exposure jobs than Canadian-born individuals. One goal of economic immigration programs is to fill labour and skills shortages. However, perceived labour shortages may eventually incentivize some employers to adopt AI technologies, especially if such shortages are in occupations highly exposed to AI -related job transformation.

Chart 7 start

Chart 7 end

Appendix Table A.1 (May 2016) and Appendix Table A.2 (May 2021) provide further results disaggregated by field of study, age group, gender, activity limitation status, selected census metropolitan area (CMA), racialized group, full-time or part-time status, union membership status, and whether the job can be done from home.

Exposure to AI -related job transformation varies substantially not only across fields of study but also on whether the employee has a bachelor’s degree or higher education. For example, employees who studied engineering and engineering technology or health care at a level below a bachelor’s degree were less likely to face AI -related job transformation than employees who studied the same disciplines at the bachelors’ degree or higher level. However, even with increased exposure, the majority of the latter group held jobs that were highly complementary with AI . Close to 60% of employees or more who studied mathematics and computer and information sciences—regardless of where they received their postsecondary education—were in high-exposure, low-complementarity jobs. Employees who studied construction trades and mechanic and repair trades may face relatively lower exposure to AI -related job transformation.

Employees aged 18 to 24 are overrepresented in low-exposure occupations, likely because they do not yet have the necessary experience to be employed in high-skill occupations. Core working-age employees, those aged 25 to 54 years, are generally more likely to hold jobs highly exposed to AI -related job transformation than their younger and older counterparts. But core working-age employees are also more likely to hold jobs that may be highly complementary with AI .

Slightly over one-fifth of men are employed in high-exposure, low-complementarity jobs, compared with 38% of women. This is because men are more likely to be employed in the skilled trades, which may face relatively lower exposure to AI -related job transformation. However, women (33%) are more likely than men (25%) to be employed in occupations that could be highly complementary with AI .

Occupations facing AI -related job transformation are more likely to be in large population centres. The CMAs of Ottawa–Gatineau (39%) and Toronto (37%) had proportionately more high-exposure, low-complementarity employment relative to other CMAs. But urban areas also had proportionately more jobs that could be highly complementary with AI .

Chinese (45%) and South Asian (38%) employees are more likely to hold high-exposure, low-complementarity jobs than other racialized groups. This is partly driven by their relatively higher representation in computer and information systems professions, which potentially highly exposed to AI -related job transformation and whose tasks may be replaceable by AI in the future. However, as noted earlier, these occupations could be integral to maintaining and improving the underlying AI infrastructure.

Unionized employees are almost as likely as their non-unionized counterparts to be highly exposed to AI -related job transformation. However, non-unionized employees (35%) are more likely to be in high-exposure, low-complementarity jobs than unionized employees (23%). This was largely driven by a higher share of unionized employees in health care and education occupations, which are potentially highly exposed to and complementary with AI .

The COVID-19 pandemic has led to significant increases in working from home (see, e.g. , Mehdi and Morissette [2021a] or Mehdi and Morissette [2021b]). These jobs are usually held by highly educated employees who may be more exposed to AI -related job transformation than their less educated counterparts. Just over half (51%) of employees with jobs that can be done from home were in high-exposure, low-complementarity occupations, compared with 14% of employees in jobs that cannot be done from home. Note However, 47% of the former group holds jobs that could be highly complementary with AI , compared with 14% of the latter group. How the advent of AI could affect the labour market in potential future pandemics is unclear (see, e.g. , Frenette and Morissette [2021]).

This study provides experimental estimates of the number and percentage of employees aged 18 to 64 in Canada potentially susceptible to AI -related job transformation using the C-AIOE index of Pizzinelli et al. (2023) and data from O*NET and the 2016 and 2021 censuses of population. Occupations were grouped into three distinct categories: (1) high exposure and low complementarity, (2) high exposure and high complementarity, and (3) low exposure. Being in the second group does not necessarily reduce AIOE , as workers would still need the necessary skills to be able to leverage the potential complementary benefits of AI .

On average, in May 2021, approximately 4.2 million employees (31%) in Canada were in the first group, about 3.9 million (29%) were in the second group and about 5.4 million (40%) were in the third group. This distribution was similar in May 2016. Employees in the following industries were more likely than others to be in the first group: professional, scientific and technical services; finance and insurance; and information and cultural industries. In contrast, employees in educational services, and health care and social assistance were more likely to be in the second group than other employees. Employees in industries such as accommodation and food services, manufacturing, construction, and transportation and warehousing face relatively less exposure to AI -related job transformation.

Unlike previous waves of automation, which affected routine and non-cognitive jobs, AI could affect a broader segment of the labour force than previously thought. Contrary to previous findings from the technological transformation literature, AI could transform the jobs of highly educated employees to a greater extent than those of their less educated counterparts. However, highly educated employees also hold jobs that may be highly complementary with AI . Previous labour market policy recommendations in response to the threat of automation included supporting upskilling and job transition initiatives. The findings in this article, which reflect the possible role of AI exposure and complementarity for occupations and workers in Canada, may inform future policy discussions on the topic.

The index used in this study is subjective and based on judgments regarding some current possibilities of AI . Consequently, the applicability of the index may decrease over time as AI capabilities grow and AI can perform an increasing number of tasks currently done by human workers. The index is also computed at the occupational level, implicitly assuming that tasks within a given occupation are the same across regions and worker characteristics. However, the ability to adapt and respond to changing skill demands will likely vary across worker characteristics. If tasks vary substantially across regions and worker characteristics, and if some tasks are more vulnerable to AI substitution, the index could be over- or underestimated to a certain extent. For example, computer programmers in one region who spend their work day coding may be more susceptible to AI -related job transformation if AI is proficient in writing that code. In contrast, programmers in another region who spend part of their day interacting face to face with team members may be less susceptible, assuming AI is not yet proficient in face-to-face interactions. To address this, future research could develop alternative measures of AI exposure at the worker level, similar to how Arntz, Gregory and Zierahn (2016) or Frenette and Frank (2020) estimated automation risk. Future studies could also attempt to answer the question, “What happened to workers whose jobs were exposed to AI -related job transformation?”

As AI technologies continue to evolve, they have the potential to reshape industries, redefine job roles and transform the nature of work. AI may also create new challenges and divides and push boundaries. But large-scale AI adoption may take some time, as employers may face financial, legal and institutional constraints. This study provides a static picture of AIOE based on employment compositions in Canada in May 2016 and May 2021, which were fairly similar. How AI affects productivity and how workers and firms adapt to the potentially evolving labour market in the long run remain to be seen.

Acemoglu, D. 2024. The Simple Macroeconomics of AI . NBER, Working Paper no. 32487.

Acemoglu, D. and S. Johnson. 2024. Learning from Ricardo and Thompson: Machinery and Labor in the Early Industrial Revolution, and in the Age of AI . NBER, Working Paper no. 32416.

Arntz, M., T. Gregory and U. Zierahn. 2016. The Risk of Automation for Jobs in OECD Countries: A Comparative Analysis . OECD Social, Employment and Migration Working Papers, no. 189. Paris: OECD Publishing.

Autor, D., H. Levy and R. Murnane. 2003. The skill content of recent technological change: An empirical exploration . The Quarterly Journal of Economics 118 (4): 1279–1333.

Beaudry, P., D. A. Green and B. M. Sand. 2016. The Great Reversal in the Demand for Skill and Cognitive Tasks . Journal of Labor Economics 34 (s1): S199–S247.

Broussard, M. 2018. Artificial Unintelligence: How Computers Misunderstand the World . Cambridge: MIT Press.

Bryan, V., Sood, S. and C. Johnston. 2024. Analysis on artificial intelligence use by businesses in Canada, second quarter of 2024 . Analysis in Brief. Statistics Canada Catalogue no. 11-621-M. Ottawa: Statistics Canada.

Cazzaniga, M., Jaumotte, F., Li, L., Melina, G., Panton, A. J., Pizzinelli, C., Rockall, E. J. and M. M. Tavares. 2024. Gen-AI: Artificial intelligence and the future of work . IMF , Working Paper no. 1.

Dingel, J. I. and B. Neiman. 2020. How many jobs can be done at home? Journal of Public Economics 189: 104235.

Eloundou, T., Manning, S., Mishkin, P. and D. Rock. 2023. GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models . Stanford Digital Economy Lab, Working Paper.

Feigenbaum, J. J. and D. P. Gross. 2023. Organizational and Economic Obstacles to Automation: A Cautionary Tale from AT&T in the Twentieth Century . NBER, Working Paper no. 29580.

Felten, E., Raj, M. and R. Seamans. 2021. Occupational, Industry, and Geographic Exposure to Artificial Intelligence: A Novel Dataset and its Potential Uses . Strategic Management Journal 42(12): 2195-2217.

Felten, E., Raj, M. and R. Seamans. 2023. How will Language Modelers like ChatGPT Affect Occupations and Industries? SSRN Working Paper.

Frenette, M. and K. Frank. 2020. Automation and Job Transformation in Canada: Who’s at Risk? Analytical Studies Branch Research Paper Series. Statistics Canada Catalogue no. 11F0019M. Ottawa: Statistics Canada.

Frenette, M. and R. Morissette. 2021. Job security in the age of artificial intelligence and potential pandemics . Economic and Social Reports (June). Ottawa: Statistics Canada.

Frey, C. B. and M. A. Osborne. 2013. The Future of Employment: How Susceptible Are Jobs to Computerisation? Oxford Martin Programme on the Impacts of Future Technology. Oxford: Oxford Martin School, University of Oxford.

Georgieff, A. and R. Hyee. 2021. Artificial intelligence and employment: New cross-country evidence . OECD Social, Employment and Migration Working Papers, no. 265. Paris: OECD Publishing.

Georgieff, A. and A. Milanez. 2021. What happened to jobs at high risk of automation? OECD Social, Employment and Migration Working Papers, no. 255. Paris: OECD Publishing.

Graetz, G. and G. Michaels. 2018. Robots at work . The Review of Economics and Statistics 100 (5): 753–768.

Kochhar, R. 2023. Which U.S. Workers Are More Exposed to AI on Their Jobs? Pew Research Center.

McElheran, K. Li, J. F., Brynjolfsson, E., Kroff, Z., Dinlersoz, E., Foster, L. S. and N. Zolas. 2024. AI Adoption in America: Who, What, and Where . NBER, Working Paper no. 31788.

Mehdi, T. and R. Morissette. 2021a. Working from home: Productivity and preferences . StatCan COVID-19: Data to Insights for a Better Canada. Statistics Canada Catalogue no. 45280001. Ottawa: Statistics Canada.

Mehdi, T. and R. Morissette. 2021b. Working from home in Canada: What have we learned so far? Economic and Social Reports (October). Ottawa: Statistics Canada.

Nedelkoska, L. and G. Quintini. 2018. Automation, Skills Use and Training . OECD Social, Employment and Migration Working Papers, no. 202. Paris: OECD Publishing.

OECD. 2023. OECD Employment Outlook 2023: Artificial Intelligence and the Labour Market . Paris: OECD Publishing.

Oschinski, M. and R. Wyonch. 2017. Future Shock? The Impact of Automation on Canada’s Labour Market . C.D. Howe Institute Commentary, no. 472. Toronto: C.D. Howe Institute.

Picot, G. and T. Mehdi. Forthcoming. The provision of high and low skilled immigrant labour to the Canadian economy .

Pizzinelli, C., Panton, A. J., Tavares, M. M., Cazzaniga, M. and L. Li. 2023. Labour market exposure to AI : Cross-country differences and distributional implications . IMF , Staff Discussion Notes no. 216.

Svanberg, M. S., Li, W., Fleming, M., Goehring, B. C. and N. C. Thompson. 2024. Beyond AI Exposure: Which Tasks are Cost-Effective to Automate with Computer Vision? MIT FutureTech, Working Paper.

Webb, M. 2020. The Impact of Artificial Intelligence on the Labor Market . SSRN, Working Paper.

Wootton, C. W. and B. E. Kemmerer. 2007. The Emergence of Mechanical Accounting in the U.S., 1880-1930 . Accounting Historians Journal 34(1): 91-124.

More information

Note of appreciation.

Canada owes the success of its statistical system to a long-standing partnership between Statistics Canada, the citizens of Canada, its businesses, governments and other institutions. Accurate and timely statistical information could not be produced without their continued co-operation and goodwill.

Standards of service to the public

Statistics Canada is committed to serving its clients in a prompt, reliable and courteous manner. To this end, the Agency has developed standards of service which its employees observe in serving its clients.

Published by authority of the Minister responsible for Statistics Canada.

© His Majesty the King in Right of Canada, as represented by the Minister of Industry, 2024

Use of this publication is governed by the Statistics Canada Open Licence Agreement .

Catalogue no.  11F0019M

Frequency: Occasional

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 02 September 2024

STAT3 promotes NLRP3 inflammasome activation by mediating NLRP3 mitochondrial translocation

• Ling Luo 1   na1 ,
• Fupeng Wang 1   na1 ,
• Xueming Xu 1 ,
• Mingliang Ma 1 ,
• Guangyan Kuang 1 ,
• Yening Zhang 1 ,
• Dan Wang 2 ,
• Ningjie Zhang 4 &
• Kai Zhao   ORCID: orcid.org/0000-0002-3889-2118 1 , 5  

Experimental & Molecular Medicine ( 2024 ) Cite this article

Metrics details

• Cell death and immune response
• Cell signalling

Recognition of the translocation of NLRP3 to various organelles has provided new insights for understanding how the NLRP3 inflammasome is activated by different stimuli. Mitochondria have already been demonstrated to be the site of NLRP3 inflammasome activation, and the latest research suggests that NLRP3 is first recruited to mitochondria, then disassociated, and subsequently recruited to the Golgi network. Although some mitochondrial factors have been found to contribute to the recruitment of NLRP3 to mitochondria, the detailed process of NLRP3 mitochondrial translocation remains unclear. Here, we identify a previously unknown role for Signal transducer and activator of transcription-3 (STAT3) in facilitating the translocation of NLRP3 to mitochondria. STAT3 interacts with NLRP3 and undergoes phosphorylation at Ser727 in response to several NLRP3 agonists, enabling the translocation of STAT3 and thus the bound NLRP3 to mitochondria. Disruption of the interaction between STAT3 and NLRP3 impairs the mitochondrial localization of NLRP3, specifically suppressing NLRP3 inflammasome activation both in vitro and in vivo. In summary, we demonstrate that STAT3 acts as a transporter for mitochondrial translocation of NLRP3 and provide new insight into the spatial regulation of NLRP3.

Similar content being viewed by others

GSK3β mediates the spatiotemporal dynamics of NLRP3 inflammasome activation

NLRP3 phosphorylation in its LRR domain critically regulates inflammasome assembly

TRIM28 SUMOylates and stabilizes NLRP3 to facilitate inflammasome activation

Introduction.

As the first line of host defense, the innate immune system uses pattern recognition receptors (PRRs) to detect invading pathogens and endogenous cellular damage to activate signaling pathways and maintain homeostasis. Numerous PRRs, including Toll-like receptors (TLRs), RIG‑I‑like receptors (RLRs) and NOD-like receptors (NLRs) 1 , 2 , 3 , have been identified. NLRs are cytosolic PRRs that induce immune responses by forming inflammasomes. Inflammasomes are supramolecular complexes consisting of sensors (NLRs), an adaptor (apoptosis-associated speck-like protein containing a CARD (ASC)) and an executor (caspase-1). Upon activation, the sensors oligomerize and then recruit ASC, which in turn recruits caspase-1 to enable its maturation. Mature caspase-1 is able to convert IL-1 family proteins into their activated forms and cleave gasdermin D (GSDMD) to trigger pyroptosis, which is responsible for inflammatory responses 4 , 5 , 6 , 7 .

The most well-studied inflammasome is the NLRP3 inflammasome. The NLRP3 inflammasome can be activated by diverse stimuli, including components of pathogens, environmental particles, and endogenous damage signals. The NLRP3 inflammasome promotes host defense against infections; however, its aberrant activation leads to several metabolic- and aging-associated inflammatory disorders, such as atherosclerosis, gout, diabetes, and Alzheimer’s disease. Notably, NLRP3 gain-of-function mutations cause autoinflammatory cryopyrin-associated periodic syndrome (CAPS). Thus, the NLRP3 inflammasome has attracted much attention because it is a highly relevant target for therapeutic intervention 8 , 9 , 10 .

NLRP3 inflammasome activation is accomplished through a two-step process comprising a priming step and an activation (or assembly) step. The priming signal is provided by pathogen ligands, such as Lipopolysaccharides (LPS), and inflammatory cytokines, such as TNF-α, which can greatly upregulate NLRP3 and IL-1β expression. Priming also mediates posttranslational modifications (PTMs) on NLRP3, which positively or negatively regulate the inflammasome. The activation signal is provided by different stimuli, such as ATP, nigericin, or monosodium urate (MSU), triggering the formation and full activation of the NLRP3 inflammasome 11 , 12 . Despite several proposed explanations, including potassium efflux, lysosomal disruption, ROS production, metabolic changes and trans-Golgi disassembly, the mechanism through which NLRP3 senses numerous stimuli remains unclear. Recently, the observation of the recruitment of NLRP3 to the Golgi and endosomal network, as well as its previously described recruitment to mitochondria and the endoplasmic reticulum (ER), have provided deep insight into how NLRP3 is activated; that is, NLRP3 needs to be in the “right place” for activation 12 . Via experiments with a live-cell multispectral time-lapse tracking system, a recent study further demonstrated that NLRP3 is translocated to mitochondria at approximately 10–15 min post stimulation 13 and is then disassociated from mitochondria and subsequently recruited to the Golgi network. This work, in addition to previous studies, have demonstrated that mitochondrial localization is required for NLRP3 activation. Although three factors (cardiolipin, mitochondrial antiviral signaling protein (MAVS), and mitofusin-2) 14 , 15 , 16 on the surface of mitochondria have been suggested to bind NLRP3 and recruit it to mitochondria, the process by which NLRP3 is transported to mitochondria is not clear.

To clarify the underlying mechanism, we investigated a series of endogenous cellular signals by screening small compounds. Surprisingly, we observed that treatment with an inhibitor of STAT3 decreased the mitochondrial localization of NLRP3 and suppressed NLRP3 inflammasome activation both in vitro and in vivo. STAT3 is a transcription factor that mediates numerous acute and chronic inflammatory processes and has a noncanonical role in regulating the function of the mitochondrial electron transport chain (ETC) 17 , 18 , 19 . Here, we demonstrated that STAT3 promotes the translocation of NLRP3 to mitochondria and its subsequent activation, revealing a new role for STAT3 in the spatial regulation of NLRP3.

Materials and Methods

Wild-type C57BL/6 mice (6–8 weeks old) were purchased from Hunan SJA Laboratory Animal Co., Ltd. (Changsha, China). All the animals were housed under specific pathogen-free (SPF) conditions in the Central South University Animal Facility. The animal experiments were conducted in accordance with the Institutional Animal Care and Use Committee of Central South University.

Small molecule compounds were purchased from Selleck Co. (Cherry Pick Library 96-well-L2000-Z451886-30 uL). The anti-caspase-1 antibody (Abcam, ab179515), anti-IL-1β antibody (RD Systems, AF-401-NA), anti-NLRP3 antibody (AdipoGen, AG-20B-0014-C100), anti-ASC antibody (AdipoGen, AG-25B-0006-C100), anti-β-actin antibody (Cell Signaling Technology, #3700), anti-STAT3 antibody (Cell Signaling Technology, #4904), anti-pY705-STAT3 antibody (Cell Signaling Technology, #9145), anti-pS727-STAT3 antibody (Cell Signaling Technology, #9134), anti-GAPDH antibody (Cell Signaling Technology, #2118), anti-VDAC antibody (Cell Signaling Technology, #4866), anti-LaminA/C antibody (Cell Signaling Technology, #4777), anti-DYKDDDDK tag antibody (Cell Signaling Technology, #2368), and anti-Myc-Tag Antibody (Cell Signaling Technology, #2272) were obtained from the indicated suppliers. Sheep anti-rabbit IgG-h + l DyLight 488 conjugate (BETHY, A120-100D2), Alexa Fluor 594 goat anti-mouse IgG (Biolegend, 405326), and protein A/G-agarose beads (Santa Cruz, sc-2003) were obtained from the indicated suppliers. Anti-Flag affinity gel (Sigma, A2220) and Pierce Anti-c-Myc agarose (Thermo Fisher Scientific, 20168) were obtained from the indicated suppliers. MitoTracker Red CMXRos (Invitrogen, M7512), MitoTracker Deep Red FM (Invitrogen, M22426), a Duolink In Situ Detection Reagents Red Kit (Sigma, DUO92008), Duolink In Situ PLA Probe Anti-Mouse MINUS (Sigma, DUO82004), and Duolink In Situ PLA Probe Anti-Rabbit PLUS (Sigma, DUO82002) were obtained from the indicated suppliers. Cell lysis buffer (Cell Signaling Technology, 9803 S), BCA Protein Assay Kits (Thermo Scientific, 23225), and an LDH Cytotoxicity Assay Kit (Beyotime, C0017) were obtained from the indicated suppliers.

Cell culture and stimulation of macrophages

Wild-type C57BL/6 mice were injected intraperitoneally with 3% thioglycollate, and three days later, primary macrophages were harvested in RPMI 1640 medium by peritoneal lavage; the purity of the isolated primary macrophages was as high as 95%, as determined by flow cytometry (Supplementary Fig. 1 ). Primary peritoneal macrophages were seeded into 6-well plates, 24-well plates, or 48-well plates depending on the experiment.

Different agonists and stimuli were used in this study as previously described in ref. 20 . For NLRP3 inflammasome activation, peritoneal macrophages were primed with LPS (100 ng/mL) for 3 h and then stimulated with ATP (5 mM, 1 h), nigericin (10 μM, 1 h) or MSU (200 μg/mL, 6 h). For AIM2 inflammasome activation, after priming, cells were transfected with poly(dA:dT) (1 μg/mL) using Lipofectamine 3000. For NLRC4 inflammasome activation, primed macrophages were transfected with Flagellin (2 μg/mL) with Lipofectamine 3000 for 1 h.

HEK293T cells were obtained from the American Type Culture Collection (Manassas, VA) and cultured in DMEM supplemented with 10% fetal bovine serum, 100 U/mL penicillin and 100 μg/mL streptomycin.

Plasmids and transfection

The NLRP3, caspase-1, pro-IL-1β, ASC and NEK7 plasmids were constructed as previously described in ref. 20 . The full-length sequence of STAT3 was amplified from iBMDM cDNA, and the sequences of the primers used were as follows: 5′-AACGGGCCCTCTAGACTCGAG ATGGCTCAGTGGAACCAGCTGCAGCAGCTGGA-3′ and 3′-TAGTCCAGTGTGGTGGAATTC CATGGGGGAGGTAGCACACTCCGAGGTCAGAT-5′. Then, the sequence was cloned and inserted into pcDNA3.1 vectors that contained different tags. Then, the plasmids were transiently transfected into HEK293T cells with linear polyethylenimine at a 1:3 mass:volume ratio. The cell culture medium was changed 6–8 h after transfection, and culture was continued. Then, the cells were collected 18–24 h later for Western blot analysis.

Immunoprecipitation and western blot analysis

Peritoneal macrophages stimulated as indicated or transfected HEK293T cells were lysed in cold IP buffer containing 50 mM Tris HCl (pH 7.4), 50 mM EDTA, 200 mM NaCl, 1% NP-40 and a protease inhibitor cocktail (Roche, 11873580001). The macrophage lysate was incubated with a primary antibody at 4 °C overnight to allow the formation of antigen-antibody complexes. The formed antigen-antibody complexes were coincubated with protein A/G-agarose beads at 4 °C for 2 h, washed three times with IP buffer, and then eluted in loading buffer. The HEK293T cell lysates were directly incubated with anti-Flag affinity gel or anti-c-Myc agarose at 4 °C for 2 h, after which the gel/agarose was washed 5 times with IP buffer and the complexes were eluted in loading buffer. The eluted samples were subjected to immunoblot analysis.

For immunoblot analysis, stimulated macrophages were lysed in cell lysis buffer supplemented with protease inhibitor cocktail and the phosphatase inhibitor PMSF. The protein concentration after lysis was determined by a BCA protein assay kit. The quantified proteins were separated by sodium dodecyl sulfate‒polyacrylamide gel electrophoresis and then transferred onto a 0.2 µM nitrocellulose membrane for immunoblot analysis.

Small interfering RNA transfection

For siRNA-mediated silencing of STAT3, cells were cultured in 24-well plates (2 × 10 5 cells per well) or 6-well plates (9 × 10 5 cells per well), and the siRNAs were then transfected with RNAiMAX Transfection Reagent (Invitrogen, 13778) following the manufacturer’s instructions. Seventy-two hours after transfection, the cells were stimulated with different inflammasome agonists. The siRNA target sequence was 5′- GCUGAAAUCAUCAUGGGCUAUTT -3′, and the scrambled negative control siRNA sequence was 5′- UUCUCCGAACGUGUCACGU-3′. The silencing efficiency was examined by western blotting using the corresponding antibodies. The indicated scrambled siRNAs were chemically synthesized by Sangon Biotech (Shanghai) Co., Ltd.

Quantitative PCR

RNA was extracted using an RNA Fast 200 Kit (Fastagen, 22001). Complementary DNA was synthesized by using TransScript All-in-One First-Strand cDNA Synthesis SuperMix for qPCR (TransGen Biotech) according to the manufacturer’s protocols. Quantitative PCR was performed using SYBR Green (Vazyme Biotech) on a LightCycler 480 instrument (Roche Diagnostics), and the expression level of each target mRNA was individually normalized to that of β-actin. The sequences of the q-PCR primers used in this study are listed in the table below:

ASC speck formation assay

To evaluate ASC speck formation, peritoneal macrophages were seeded on chamber slides and allowed to attach overnight. The following day, the cells were primed with LPS and treated with nigericin in the presence or absence of the indicated inhibitors. Then, the cells were fixed with 4% paraformaldehyde in PBS for 15 min and washed three times in PBS with Tween 20 (PBST) prior to permeabilization with 0.1% Triton X-100 for 10 min. After blocking with 3% bovine serum albumin in PBS for 1 h, the cells were incubated with primary antibodies overnight at 4 °C. After washing with PBST, the cells were incubated with secondary antibodies in 3% bovine serum albumin for 30 min, and nuclei were then stained with DAPI (Beyotime, P0131). The cells were visualized by fluorescence microscopy (Nikon Ti2-U).

Proximity ligation assay

Proximity ligation assays were performed using Duolink reagents (Sigma) according to the manufacturer’s instructions to visualize the interaction between the STAT3 and NLRP3 proteins and their localization in mitochondria-resident mouse peritoneal macrophages. To study the interaction between STAT3 and NLRP3, cells were grown on PTFE printed microscope slides (Electron Microscopy Science, 63423-08) and subjected to canonical stimulation of the NLRP3 pathway in the presence or absence of the mitochondrial dye MitoTracker Deep Red FM. The cells were fixed, permeabilized, blocked, and then incubated with primary antibodies overnight at 4 °C. After incubation with the primary antibodies, the cells were incubated with a combination of the corresponding PLA probes and secondary antibodies conjugated to oligonucleotides for 1 h at 37 °C. After washing with buffer A [0.01 M Tris-HCl (pH 7.4), 0.15 M NaCl, 0.05% Tween 20], the cells were incubated with ligation mix (Sigma, DUO92008) for 30 min at 37 °C to allow the formation of a closed circle DNA template when the PLA probes were bound in close proximity. After washing with buffer A, the cells were incubated with polymerase mix (Sigma, DUO92008) for 100 min at 37 °C to allow rolling circle amplification. After sequential washes with buffer B [0.2 M Tris-HCl (pH 7.4), 0.1 M NaCl] and buffer C (10-fold dilution of buffer B with water), cover slips (Citotest, 10212450 C) were mounted onto the microscopy slides. The cells were imaged using a 63×/1.4 oil immersion objective on a Leica STELLARIS 5 confocal microscope (Leica).

Immunofluorescence staining and confocal microscopy

Primed cells were treated with DMSO or with napabucasin followed by nigericin for 30 min. At the same time, the cells were stained for 30 min at 37 °C with 100 nM MitoTracker Red CMXRos (Invitrogen, M7512) in the dark. After washing three times with warm PBS, the cells were fixed with 4% paraformaldehyde in PBS for 15 min and permeabilized with 0.1% Triton X-100 for 10 min. After blocking with 3% bovine serum albumin in PBS for 1 h, the cells were incubated overnight at 4 °C with primary antibodies (anti-NLRP3 and anti-STAT3 antibodies). After washing with PBST, the cells were incubated with Sheep anti-rabbit IgG-h + l DyLight 488 conjugate and Alexa Fluor 594-conjugated goat anti-mouse IgG in PBS containing 3% BSA for 30 min and rinsed in PBST. Nuclei were stained with DAPI (Beyotime, P0131). Images were analyzed by using a 63×/1.4 oil immersion objective on a Leica STELLARIS 5 confocal microscope (Leica).

Cellular thermal shift assay (CETSA)

Peritoneal macrophages were collected and resuspended at a density of 4 × 10 5  cells/mL in a 1.5 mL Eppendorf tube and were then washed with ice-cold PBS. The macrophages were resuspended in RIPA lysis buffer containing the protease inhibitor cocktail. The cell lysates were centrifuged at 12,000 × g for 15 min at 4 °C, after which the supernatants were collected. The lysate supernatants were incubated with napabucasin (20 μM) or the control (DMSO) for 1 h at room temperature. Then, each drug-treated cell lysates was aliquoted into PCR tubes (60 μL each), and the tubes were heated at different temperatures (37, 47, 57, 67, 77, and 87 °C) for 5 min in the PCR apparatus and then cooled at room temperature for 3 min. The soluble fractions were separated after centrifugation of the lysates at 20,000 × g for 20 min at 4 °C, transferred to new microcentrifuge tubes, and analyzed by SDS‒PAGE followed by western blotting with anti-NLRP3 and anti-STAT3 antibodies.

Separation of the cytoplasmic and nuclear fractions

The cytoplasmic and nuclear fractions were separated from lysates of peritoneal macrophages using a cytoplasmic and nuclear fractionation kit (Beyotime, P0028) according to the manufacturer’s guidelines. Then, the isolated protein components were quantified by a BCA assay and analyzed by immunoblotting.

ELISA and LDH assay

Supernatants from cell cultures and sera were analyzed using IL-1β and TNF-a ELISA kits (Invitrogen, 88-5019-88, 88-7324-88) according to the manufacturer’s instructions. Cell death was assessed with an LDH Cytotoxicity Assay Kit (Beyotime, C0017) according to the manufacturer’s instructions.

In vivo LPS challenge

Wild-type C57BL/6 male mice were injected with saline or napabucasin (5 mg/kg) half an hour before injection with LPS (20 mg/kg body weight) and were then monitored for 8 h. The mice were deeply anesthetized with sodium pentobarbital (70 mg/kg, i.p.), and the cardiac blood and left lung were collected from each mouse. Then, the serum concentrations of IL-1β and TNF-α were measured by ELISA, and lung damage was evaluated by HE staining.

MSU-induced peritonitis modeling in vivo

Wild-type C57BL/6 male mice were intraperitoneally injected with saline or napabucasin (5 mg/kg), half an hour before injection of 2 mg of MSU (dissolved in 300 μL of saline) and were then monitored for 6 h. After the mice were killed, the peritoneal lavage fluid was collected by lavage of the peritoneal cavity with ice-cold PBS; the lavage fluid was concentrated for ELISA, and neutrophils were counted.

Statistical analysis

The data were analyzed using GraphPad Prism software (version 9.5.0). The data are presented as the standard deviation of the mean (SD) or standard error of the mean (SEM), depending on the experiment. Independent sample t-tests and two-way ANOVA followed by the Bonferroni correction were performed in this study. Differences with a p -value < 0.05 were considered to be statistically significant, and the details of the statistical analyses can be found in the figure legends. In the animal studies, the mice were randomly divided into different groups, and all the samples were processed in a blinded manner. For western blotting, qPCR and other quantitative methods, experiments were performed at least three times.

Identification of STAT3 inhibitors that regulate the NLRP3 inflammasome

To elucidate whether endogenous cellular signals are involved in NLRP3 translocation, we detected IL-1β secreted in response to treatment with NLRP3 agonists (LPS+nigericin) and screened 395 small compounds obtained from Selleck Co. (Supplementary Table 1 ) for their ability to inhibit IL-1β secretion (Fig. 1a ). These compounds were inducers or inhibitors of numerous intrinsic signaling pathways, including the MAPK, PI3K/AKT/mTOR, JAK-STAT, TGF-β-Smad, and integrin signaling pathways. We treated macrophages with the compounds after treatment with LPS to focus on regulation at the NLRP3 inflammasome activation stage rather than the priming stage (Fig. 1a ). Intriguingly, 20 compounds inhibited IL-1β secretion by at least 90%, and three of them, namely, napabucasin, BP-1-102 and C188-9, attracted our attention (Fig. 1b ) since they are inhibitors of STAT3.

a Screening for small molecule compounds obtained from Selleck Co. This figure template, “Screening flow chart,” was assembled using dynamic BioRender assets (icons, lines, shapes and/or text) and is fully editable. b Percentage inhibition of IL-1β secretion by each compound. The IL-1β secretion level relative to 10% of that in the control group was used as a cutoff (left), and the chemical structures of the compounds napabucasin, BP-1-102 and C188-9 used to inhibit STAT3 are shown (right).

STAT3 promotes NLRP3 inflammasome activation in both transcription-dependent and transcription-independent manners

To further confirm the function of STAT3 inhibitors in inflammasome activation, we treated macrophages with nigericin, ATP and MSU, which activate the NLRP3 inflammasome via different mechanisms, as well as with flagellin and poly(dA:dT), which activate the NLRC4 and AIM2 inflammasomes, respectively. Given that napabucasin has been applied in clinical trials, we chose to use napabucasin for follow-up experiments 21 , 22 . We observed that napabucasin greatly inhibited IL-1β secretion but not TNF-α secretion in response to stimulation of the NLRP3 inflammasome, although it had no effect on the NLRC4 or AIM2 inflammasome. Accordingly, lactic acid dehydrogenase (LDH) release triggered by NLRP3 stimulation was also impaired by napabucasin treatment (Fig. 2a ). This observation was also verified by the detection of cleaved caspase-1 (p10) and pro-IL-1β (p17) in the supernatants of macrophages treated with these agonists (Fig. 2b, c ). We also noted that napabucasin barely affected the protein levels of NLRP3 and pro-IL-1β after LPS priming (Fig. 2b, c ), excluding its role in the regulation of the NF-κB pathway. ASC speck formation is a hallmark of NLRP3 inflammasome assembly and can be monitored by immunofluorescence staining. We further observed that napabucasin inhibited ASC speck formation in ATP- and nigericin-treated macrophages but not in flagellin- or poly(dA:dT)-treated macrophages (Fig. 2d ). Thus, these results demonstrate that STAT3 inhibitors specifically suppress the activation of the NLRP3 inflammasome, suggesting that STAT3 may be involved in the regulation of NLRP3 inflammasome assembly via an unknown mechanism.

a ELISA of IL-1β and TNF-α secretion and assay of LDH release in supernatants from LPS-primed mouse peritoneal macrophages treated with 10 μM napabucasin for 30 min followed by the indicated stimulators of different inflammasomes. The experiments were repeated at least three times, and representative data are shown. b , c Immunoblot analysis of supernatants (SN) or cell lysates (Cell) from mouse peritoneal macrophages treated with the indicated stimulators of the NLRP3 ( b ), NLRC4 or AIM2 inflammasome with or without napabucasin ( c ). d Representative images of ASC specks in peritoneal macrophages treated with the indicated stimuli; ASC, green; nuclei, blue. The white arrows indicate ASC specks. Scale bar: 20 μm. The percentage of cells containing an ASC speck was quantified (right). At least 100 peritoneal macrophages from each genotype were analyzed. The results are presented as the means ± SDs, and representative photographs from three biologically independent experiments with similar results are shown. Statistical analyses were carried out via two-way ANOVA with the Bonferroni correction in ( a , d ). * P  < 0.05, ** P  < 0.01 and *** P  < 0.001.

Next, we examined the role of STAT3 in the regulation of the NLRP3 inflammasome. STAT3 was knocked down in mouse primary macrophages by siRNA transfection (Fig. 3a ), and silencing STAT3 indeed suppressed IL-1β secretion in macrophages treated with NLRP3, NLRC4, or AIM2 agonists (Fig. 3b ). We found that silencing STAT3 also decreased NLRP3 and pro-IL-1β expression at the translational and transcriptional levels 23 , 24 , 25 (Fig. 3c, d ), in contrast to the results obtained with the STAT3 inhibitor. These effects of STAT3 silencing were consistent with those observed in previous studies indicating that STAT3 could promote NLRP3 and IL-1β expression at the transcriptional level. Taken together, these findings indicate that either knockdown of STAT3 or blockade of STAT3 by treatment with inhibitors can suppress NLRP3 inflammasome activation via different mechanisms.

a Quantitative PCR analysis of STAT3 mRNA expression in mouse peritoneal macrophages after transfection with NC-siRNA or STAT3-siRNA for 48 h. b IL-1β and TNF-α secretion in supernatants from mouse peritoneal macrophages transfected as described in ( a ) and then treated with the indicated stimuli. c , d Relative NLRP3 and IL-1β mRNA expression ( c ) in peritoneal macrophages treated with LPS for 2 h and 4 h or not treated after transfection as described in ( a ). Target mRNA expression was normalized to the expression of β-actin as the reference gene. Immunoblot analysis of NLRP3 and Pro-IL-1β expression in mouse peritoneal macrophages transfected as described; the photographs are representative of three biologically independent experiments with similar results ( d ). The results are presented as the means ± SDs; n  = 3 biologically independent experiments ( a – c ). Statistical analyses were carried out via independent sample t-test in ( a ) or two-way ANOVA with the Bonferroni correction in ( b , c ). * P  < 0.05, ** P  < 0.01 and *** P  < 0.001.

Cytoplasmic STAT3 interacts with NLRP3

To further study the nontranscriptional function of STAT3 in NLRP3 inflammasome activation, we examined the distribution of STAT3 during the process of NLRP3 inflammasome activation. STAT3 translocated to the nucleus upon LPS treatment (Fig. 4a ), as suggested by a previous study. Intriguingly, upon the addition of nigericin or ATP, STAT3 was located primarily in the cytoplasm (Fig. 4a ), suggesting that cytoplasmic STAT3 may play a role in the assembly of the NLRP3 inflammasome. Therefore, we investigated whether cytoplasmic STAT3 can interact with NLRP3 inflammasome components. By overexpressing STAT3 with NLRP3, ASC, NEK7 or caspase-1 in HEK293T cells, we found that STAT3 could interact with NLRP3 but not with ASC, NEK7 or caspase-1 (Fig. 4b ). This interaction was verified by immunoprecipitation in primary macrophages, and nigericin treatment dramatically promoted the association between STAT3 and NLRP3 (Fig. 4c ). A PLA (proximity ligation assay) assay was further used to demonstrate this interaction in primary macrophages (Fig. 4d ). Moreover, to determine which domains of STAT3 and NLRP3 are required for this interaction, we constructed plasmids expressing different truncation mutants of STAT3 and NLRP3. We observed that the DBD of STAT3 and at least two domains, the PYD and LRR domains, of NLRP3 mediate the interaction between STAT3 and NLRP3 (Fig. 4e, f ). Thus, cytoplasmic STAT3 interacts with NLRP3 at the NLRP3 inflammasome activation stage.

a Immunoblot analysis of cytoplasmic and nuclear components after treatment with the corresponding stimuli described above. b Immunoprecipitation and immunoblot analysis of lysates from HEK293T cells transfected with Flag-STAT3, Myc-NLRP3, Myc-ASC, Myc-STAT3, Flag-caspase-1, or Flag-Nek7. Immunoprecipitation was performed with anti-Flag (up) and anti-Myc (down) antibodies, and immunoblotting was performed with anti-Myc and anti-Flag antibodies, respectively. c Immunoblot analysis of lysates from mouse peritoneal macrophages treated with the indicated stimuli. IP was performed with an anti-STAT3 antibody, and immunoblotting was then performed with an anti-NLRP3 antibody. d The physical interaction between NLRP3 and STAT3 was visualized as red puncta by a PLA in mouse peritoneal macrophages primed with 100 ng/mL LPS for 3 h and then stimulated with nigericin (10 μM) or ATP (5 mM) for 1 h. Scale bar: 20 μm. e Myc-tagged NLRP3 or its mutants and Flag-STAT3 were cotransfected into HEK293T cells for 24 h prior to immunoprecipitation with anti-Myc beads and western blotting (bottom). Schematic diagram of NLRP3 and its truncation mutants (top). f Myc-STAT3 or its mutants and Flag-NLRP3 were cotransfected into HEK293T cells for 24 h prior to immunoprecipitation with anti-Myc beads and western blotting (bottom). Schematic diagram of STAT3 and its truncation mutants (top). The results are presented as the means ± SDs, and representative photographs of three biologically independent experiments with similar results are shown.

STAT3 transports NLRP3 to mitochondria

Then, we further investigated the mechanism by which cytoplasmic STAT3 regulates NLRP3 inflammasome activation. Since STAT3 has been reported to regulate ETC function in mitochondria 19 , 26 , we speculated that the mitochondrial function of STAT3 contributes to NLRP3 inflammasome activation. We observed that treatment with the NLRP3 agonists nigericin and ATP increased Ser727 phosphorylation of STAT3 (Fig. 5a ), which is a marker for STAT3 translocation to mitochondria 19 , 23 . Intriguingly, the STAT3-NLRP3 complex was colocalized with a mitochondrial marker (MitoTracker) (Fig. 5b ). Thus, these results indicate that STAT3 can transport NLRP3 to mitochondria, as the translocation of NLRP3 to mitochondria has been reported to play a pivotal role in its activation. To verify this hypothesis, we sought to disrupt the interaction between STAT3 and NLRP3. Given that knockdown of STAT3 suppresses the expression of NLRP3, we cannot exclude the transcriptional effect of STAT3 on the regulation of NLRP3 inflammasome. Instead, we used napabucasin for the mechanistic study because it has been reported to bind to the SH2 domain of STAT3 and block STAT3 activity 21 , 27 . We first evaluated the potential of napabucasin to bind to STAT3 by a cellular thermal shift assay (CETSA) 28 , which detects the thermal stability of a protein upon ligand binding. Napabucasin increased the thermal stability of STAT3 but not that of NLRP3 (Fig. 5c ), indicating that napabucasin indeed binds to STAT3. Thus, napabucasin was appropriate for use in subsequent mechanistic studies. The translocation of NLRP3 to mitochondria was significantly impaired in macrophages treated with napabucasin compared to control macrophages (Fig. 5d ). Furthermore, napabucasin disrupted the interaction between STAT3 and NLRP3 during the assembly of the NLRP3 inflammasome after stimulation with LPS or Pam3CSK4, which activate the TLR4 and TLR1/2 pathways, respectively (Fig. 5e , Supplementary Fig. 2 ), but did not affect Ser727 phosphorylation of STAT3 (Fig. 5f ). Taken together, these findings demonstrated that STAT3 mediates the translocation of NLRP3 to mitochondria for further activation of the NLRP3 inflammasome.

a Immunoblot analysis of cytoplasmic components using a cytoplasmic and nuclear fractionation kit after treatment with the corresponding stimuli described above. b The interaction between NLRP3 and STAT3 and the co-localization with mitochondria (red) in peritoneal macrophages were visualized by a PLA (green) and confocal microscopy. Scale bar: 10 μm. c Cellular thermal shift assay (CETSA) of STAT3 or NLRP3 with napabucasin (20 μM). d LPS-primed peritoneal macrophages were treated with the indicated stimuli. The colocalization of NLRP3 with mitochondria was visualized by immunofluorescence microscopy. Mitochondria were stained with MitoTracker CMXRos, NLRP3 was detected with Alexa Fluor 488, and cellular nuclei were stained with DAPI. The imaging data are representative of several images from three independent experiments. Scale bar: 20 μm. e Immunoprecipitation and immunoblot analysis of lysates from mouse peritoneal macrophages treated with the indicated stimuli with or without napabucasin (10 μM) prior to IP with an anti-STAT3 antibody and immunoblotting with an anti-NLRP3 antibody. f Cytoplasmic components were extracted with a cytoplasmic and nuclear extraction kit after stimulation as described above and were then analyzed by Western blotting. The results are presented as the means ± SDs, and representative photographs of three biologically independent experiments with similar results are shown.

Targeting STAT3 alleviates NLRP3-associated inflammation

Finally, we explored the potential of targeting STAT3 to alleviate NLRP3-associated inflammation. We established a model of peritonitis in mice by intraperitoneal (i.p.) injection of MSU, in which inflammation is NLRP3 inflammasome dependent. MSU challenge induced neutrophil infiltration and IL-1β secretion in peritoneal fluids, whereas napabucasin pretreatment reduced both of these parameters (Fig. 6a, b ). In another LPS-induced endotoxin model, napabucasin exhibited anti-inflammatory effects by decreasing the serum IL-1β level and reducing lung tissue damage, alveolar edema and neutrophil infiltration compared with mice those in mice only challenged with LPS (Fig. 6c–e ). Thus, targeting STAT3 could be a potential treatment strategy for NLRP3-associated inflammation.

a , b Wild-type C57BL/6 mice were administered DMSO or napabucasin (5 mg/kg) via intraperitoneal (i.p.) injection 30 min before i.p. injection of MSU (2 mg per mouse) ( n  = 5 biologically independent mice) for 6 h. Quantification of neutrophils ( a ) and ELISA of IL-1β secretion (b) in the peritoneal lavage fluid. c – e Wild-type C57BL/6 mice were administered 20 mg/kg LPS via intraperitoneal (i.p.) injection ( n  = 5 biologically independent mice) for 8 h with or without napabucasin (5 mg/kg). ELISA results showing the serum concentrations of IL-1β ( c ) and IL-6 ( d ) and images of H&E-stained lung tissue sections ( e ). Scale bar: 50 μm. The data are representative of three independent experiments. The results are presented as the means ± SEMs. Statistical analyses were carried out via two-way ANOVA in ( a – d ). * P  < 0.05, ** P  < 0.01 and *** P  < 0.001.

Accumulating evidence has revealed that mitochondria play a crucial role in NLRP3 inflammasome activation via several mechanisms, including acting as scaffolds for the localization of NLRP3, releasing mitochondrial DNA (mtDNA) and mitochondrial ROS (mtROS) into the cytoplasm, and providing ATP for NLRP3 signaling, all of which contribute to the formation and activation of the NLRP3 inflammasome 29 , 30 . Although cardiolipin, MAVS and mitofusin-2 on the surface of mitochondria have been suggested to bind NLRP3 and recruit it to mitochondria 14 , 15 , 16 , how NLRP3 is translocated to mitochondria remains largely unknown. In this study, we demonstrated that STAT3 acts as a transporter for NLRP3 translocation to mitochondria (Fig. 7 ) and that blocking the interaction between STAT3 and NLRP3 by treatment with a STAT3 inhibitor substantially reduced NLRP3 inflammasome activation both in vitro and in vivo.

STAT3 promotes NLRP3 inflammasome activation by mediating NLRP3 mitochondrial translocation, and this process is inhibited by napabucasin through disruption of the interaction between STAT3 and NLRP3.

The knowledge of the translocation of NLRP3 to various organelles represents the greatest advancement in the field. Zhijian J. Chen’s group reported that diverse NLRP3 stimuli induce disassembly of the trans-Golgi network (TGN), after which the dispersed TGN (dTGN) serves as a scaffold for NLRP3 oligomerization and activation 31 . That work provided a new direction for exploring the underlying mechanism of NLRP3 activation. Subsequent studies revealed that BTK 32 , IKKβ 33 , and GSK3β 13 are involved in regulating the location of NLRP3 in the dTGN. The most recent study showed, by using a live-cell multispectral time-lapse tracking system, that NLRP3 first translocates to mitochondria at approximately 10–15 min post stimulation and is subsequently recruited to the Golgi network 12 . This work, combined with previous studies, highlights the importance of NLRP3 translocation to mitochondria. In the present study, we also observed the translocation of NLRP3 to mitochondria upon stimulation of NLRP3, and disruption of NLRP3 mitochondrial translocation impaired the activation of the NLRP3 inflammasome. In mediating the translocation of NLRP3 to mitochondria, STAT3 mostly acts as a transporter by binding to NLRP3, which is different from the mechanism by which BTK, IKKβ and GSK3β in regulate the localization of NLRP3 to the dTGN. BTK 32 , IKKβ 33 and GSK3β 13 regulate the localization of NLRP3 to the dTGN in a manner dependent on their kinase function. For example, BTK phosphorylates specific tyrosine residues in the polybasic region of NLRP3, resulting in charge reversal in this region. This change promotes NLRP3 disassociation from the dTGN 32 . IKKβ and GSK3β have similar functions, but NLRP3 is not one of their substrates 13 , 33 . Previous studies have reported that Cardiolipin, MAVS and mitofusin-2 mediate the recruitment of NLRP3 to mitochondria by interacting with NLRP3 14 , 15 , 16 . However, these factors are located on the surface of mitochondria or the inner mitochondrial membrane; thus, how can they enable the translocation of cytosolic NLRP3 to mitochondria? Our results partially answer this question and provide a new explanation for the translocation of NLRP3 to mitochondria; i.e., that it is mediated by a transporter. Future work is needed to explore this dynamic process.

STAT3, a member of the STAT protein family, is a transcription factor that extensively participates in the regulation of acute and chronic inflammation, autoimmunity, metabolism, development and cancer progression 34 , 35 , 36 . It can be activated by various cytokines and growth factors, including IL-6, IL-10, IL-11, interferon, EGF, and HGF 37 . Upon activation, STAT3 is phosphorylated at two well-studied sites, Tyr705 and Ser727. When STAT3 undergoes phosphorylation at Tyr705, it is transported to the nucleus, where it can specifically bind to DNA for transcriptional activation 34 . The transcriptional regulation of NLRP3 and IL-1β by STAT3 has been investigated by several groups; however, no group has revealed a transcriptionally independent role for STAT3 in NLRP3 inflammasome regulation. By treatment with an inhibitor of STAT3 after LPS priming, we revealed the noncanonical function of STAT3 in NLRP3 inflammasome activation. This noncanonical function relies on Ser727 phosphorylation. When STAT3 undergoes phosphorylation at Ser727, it is transported to mitochondria and regulates ETC function 18 , 23 . Intriguingly, NLRP3 agonists can trigger Ser727 phosphorylation of STAT3; although the mechanism underlying this process is unknown, kinases responsible for Ser727 phosphorylation of STAT3 must exist. Several kinases, including EGF, PKC, JNK, ERK1, ERK2 and MAP kinases, have been reported to mediate Ser727 phosphorylation of STAT3 37 , 38 , 39 . Whether these kinases or unknown kinases are involved in this process still needs further investigation. In this study, we demonstrated that napabucasin binds to STAT3, but we cannot exclude the unknown effects of napabucasin on the NLRP3 inflammasome. In future studies, a series of STAT3 mutant mice which lost mitochondrial translocation effect needs to be investigated. In summary, we revealed a new role for STAT3 in regulating the location of NLRP3.

Cao, X. Self-regulation and cross-regulation of pattern-recognition receptor signalling in health and disease. Nat. Rev. Immunol. 16 , 35–50, https://doi.org/10.1038/nri.2015.8 (2016).

Article   CAS   PubMed   Google Scholar  

Brubaker, S. W., Bonham, K. S., Zanoni, I. & Kagan, J. C. Innate immune pattern recognition: a cell biological perspective. Annu Rev. Immunol. 33 , 257–290, https://doi.org/10.1146/annurev-immunol-032414-112240 (2015).

Article   CAS   PubMed   PubMed Central   Google Scholar  

Kawai, T. & Akira, S. The role of pattern-recognition receptors in innate immunity: update on Toll-like receptors. Nat. Immunol. 11 , 373–384, https://doi.org/10.1038/ni.1863 (2010).

Mantovani, A., Dinarello, C. A., Molgora, M. & Garlanda, C. Interleukin-1 and Related Cytokines in the Regulation of Inflammation and Immunity. Immunity 50 , 778–795, https://doi.org/10.1016/j.immuni.2019.03.012 (2019).

Broz, P., Pelegrin, P. & Shao, F. The gasdermins, a protein family executing cell death and inflammation. Nat. Rev. Immunol. 20 , 143–157, https://doi.org/10.1038/s41577-019-0228-2 (2020).

Lamkanfi, M. & Dixit, V. M. Mechanisms and functions of inflammasomes. Cell 157 , 1013–1022, https://doi.org/10.1016/j.cell.2014.04.007 (2014).

Rathinam, V. A., Vanaja, S. K. & Fitzgerald, K. A. Regulation of inflammasome signaling. Nat. Immunol. 13 , 333–342, https://doi.org/10.1038/ni.2237 (2012).

Swanson, K. V., Deng, M. & Ting, J. P. The NLRP3 inflammasome: molecular activation and regulation to therapeutics. Nat. Rev. Immunol. 19 , 477–489, https://doi.org/10.1038/s41577-019-0165-0 (2019).

Sharma, B. R. & Kanneganti, T. D. NLRP3 inflammasome in cancer and metabolic diseases. Nat. Immunol. 22 , 550–559, https://doi.org/10.1038/s41590-021-00886-5 (2021).

Mangan, M. S. J. et al. Targeting the NLRP3 inflammasome in inflammatory diseases. Nat. Rev. Drug Discov. 17 , 588–606, https://doi.org/10.1038/nrd.2018.97 (2018).

Huang, Y., Xu, W. & Zhou, R. NLRP3 inflammasome activation and cell death. Cell Mol. Immunol. 18 , 2114–2127, https://doi.org/10.1038/s41423-021-00740-6 (2021).

Akbal, A. et al. How location and cellular signaling combine to activate the NLRP3 inflammasome. Cell Mol. Immunol. 19 , 1201–1214, https://doi.org/10.1038/s41423-022-00922-w (2022).

Arumugam, S. et al. GSK3beta mediates the spatiotemporal dynamics of NLRP3 inflammasome activation. Cell Death Differ. 29 , 2060–2069, https://doi.org/10.1038/s41418-022-00997-y (2022).

Iyer, S. S. et al. Mitochondrial cardiolipin is required for Nlrp3 inflammasome activation. Immunity 39 , 311–323, https://doi.org/10.1016/j.immuni.2013.08.001 (2013).

Subramanian, N., Natarajan, K., Clatworthy, M. R., Wang, Z. & Germain, R. N. The adaptor MAVS promotes NLRP3 mitochondrial localization and inflammasome activation. Cell 153 , 348–361, https://doi.org/10.1016/j.cell.2013.02.054 (2013).

Tur, J. et al. Mitofusin 2 in Macrophages Links Mitochondrial ROS Production, Cytokine Release, Phagocytosis, Autophagy, and Bactericidal Activity. Cell Rep. 32 , 108079, https://doi.org/10.1016/j.celrep.2020.108079 (2020).

Yang, R. & Rincon, M. Mitochondrial Stat3, the Need for Design Thinking. Int J. Biol. Sci. 12 , 532–544, https://doi.org/10.7150/ijbs.15153 (2016).

Hillmer, E. J., Zhang, H., Li, H. S. & Watowich, S. S. STAT3 signaling in immunity. Cytokine Growth Factor Rev. 31 , 1–15, https://doi.org/10.1016/j.cytogfr.2016.05.001 (2016).

Article   PubMed   PubMed Central   Google Scholar  

Wegrzyn, J. et al. Function of mitochondrial Stat3 in cellular respiration. Science 323 , 793–797, https://doi.org/10.1126/science.1164551 (2009).

Wang, D. et al. YAP promotes the activation of NLRP3 inflammasome via blocking K27-linked polyubiquitination of NLRP3. Nat. Commun. 12 , 2674, https://doi.org/10.1038/s41467-021-22987-3 (2021).

Li, Y. et al. Suppression of cancer relapse and metastasis by inhibiting cancer stemness. Proc. Natl Acad. Sci. USA 112 , 1839–1844, https://doi.org/10.1073/pnas.1424171112 (2015).

Jonker, D. J. et al. Napabucasin versus placebo in refractory advanced colorectal cancer: a randomised phase 3 trial. Lancet Gastroenterol. Hepatol. 3 , 263–270, https://doi.org/10.1016/S2468-1253(18)30009-8 (2018).

Article   PubMed   Google Scholar  

Balic, J. J. et al. STAT3 serine phosphorylation is required for TLR4 metabolic reprogramming and IL-1beta expression. Nat. Commun. 11 , 3816, https://doi.org/10.1038/s41467-020-17669-5 (2020).

Samavati, L. et al. STAT3 tyrosine phosphorylation is critical for interleukin 1 beta and interleukin-6 production in response to lipopolysaccharide and live bacteria. Mol. Immunol. 46 , 1867–1877, https://doi.org/10.1016/j.molimm.2009.02.018 (2009).

Zhu, L. et al. STAT3/Mitophagy Axis Coordinates Macrophage NLRP3 Inflammasome Activation and Inflammatory Bone Loss. J. Bone Min. Res 38 , 335–353, https://doi.org/10.1002/jbmr.4756 (2023).

Article   CAS   Google Scholar  

Carbognin, E., Betto, R. M., Soriano, M. E., Smith, A. G. & Martello, G. Stat3 promotes mitochondrial transcription and oxidative respiration during maintenance and induction of naive pluripotency. EMBO J. 35 , 618–634, https://doi.org/10.15252/embj.201592629 (2016).

Tahara, T. et al. STAT3 inhibitory activity of naphthoquinones isolated from Tabebuia avellanedae. Bioorg. Med Chem. 28 , 115347, https://doi.org/10.1016/j.bmc.2020.115347 (2020).

Peng, Y. et al. Lovastatin inhibits Toll-like receptor 4 signaling in microglia by targeting its co-receptor myeloid differentiation protein 2 and attenuates neuropathic pain. Brain Behav. Immun. 82 , 432–444, https://doi.org/10.1016/j.bbi.2019.09.013 (2019).

Xian, H. et al. Oxidized DNA fragments exit mitochondria via mPTP- and VDAC-dependent channels to activate NLRP3 inflammasome and interferon signaling. Immunity 55 , 1370–1385.e1378, https://doi.org/10.1016/j.immuni.2022.06.007 (2022).

Xian, H. et al. Metformin inhibition of mitochondrial ATP and DNA synthesis abrogates NLRP3 inflammasome activation and pulmonary inflammation. Immunity 54 , 1463–1477.e1411, https://doi.org/10.1016/j.immuni.2021.05.004 (2021).

Chen, J. & Chen, Z. J. PtdIns4P on dispersed trans-Golgi network mediates NLRP3 inflammasome activation. Nature 564 , 71–76, https://doi.org/10.1038/s41586-018-0761-3 (2018).

Bittner, Z. A. et al. BTK operates a phospho-tyrosine switch to regulate NLRP3 inflammasome activity. J. Exp. Med. 218 , https://doi.org/10.1084/jem.20201656 (2021).

Nanda, S. K., Prescott, A. R., Figueras-Vadillo, C. & Cohen, P. IKKbeta is required for the formation of the NLRP3 inflammasome. EMBO Rep. 22 , e50743, https://doi.org/10.15252/embr.202050743 (2021).

Reich, N. C. & Liu, L. Tracking STAT nuclear traffic. Nat. Rev. Immunol. 6 , 602–612, https://doi.org/10.1038/nri1885 (2006).

Zou, S. et al. Targeting STAT3 in Cancer Immunotherapy. Mol. Cancer 19 , 145, https://doi.org/10.1186/s12943-020-01258-7 (2020).

Johnson, D. E., O’Keefe, R. A. & Grandis, J. R. Targeting the IL-6/JAK/STAT3 signalling axis in cancer. Nat. Rev. Clin. Oncol. 15 , 234–248, https://doi.org/10.1038/nrclinonc.2018.8 (2018).

Yu, H., Pardoll, D. & Jove, R. STATs in cancer inflammation and immunity: a leading role for STAT3. Nat. Rev. Cancer 9 , 798–809, https://doi.org/10.1038/nrc2734 (2009).

Sgrignani, J. et al. Structural Biology of STAT3 and Its Implications for Anticancer Therapies Development. Int. J. Mol. Sci. 19 , https://doi.org/10.3390/ijms19061591 (2018).

Macias, E., Rao, D., Carbajal, S., Kiguchi, K. & DiGiovanni, J. Stat3 binds to mtDNA and regulates mitochondrial gene expression in keratinocytes. J. Invest Dermatol. 134 , 1971–1980, https://doi.org/10.1038/jid.2014.68 (2014).

Download references

Acknowledgements

This work was supported by the National Key Research and Development Program of China (2021YFC2500802), the National Natural Science Foundation of China (82272207, 82102281, 82202391, 82000831), the Provincial Natural Science Foundation of Hunan in China (2021JJ20090, 2022JJ20093), the Wisdom Accumulation and Talent Cultivation Project of Third Xiangya Hospital of Central South University (JC202201, YX202201), the Graduate Student Research Innovation Project in Hunan Province (CX20230316), the Key R&D Plan of Hunan Province (NO2023SK2035), and an Openning Project of Key Laboratory of Surgical Critical Care and Life Support (Xi’an Jiaotong University), Ministry of Education (2023SCCLS-KFKT001).

Author information

These authors contributed equally: Ling Luo, Fupeng Wang.

Authors and Affiliations

Department of Hematology and Critical Care Medicine, Third Xiangya Hospital, Central South University, Changsha, Hunan Province, 410000 P, PR China

Ling Luo, Fupeng Wang, Xueming Xu, Mingliang Ma, Guangyan Kuang, Yening Zhang & Kai Zhao

Department of Dermatology, Third Xiangya Hospital, Central South University, Changsha, Hunan Province, 410000 P, PR China

Department of Rheumatology, First Affiliated Hospital of Zhengzhou University, Zhengzhou, Henan Province, 450000 P, PR China

Department of Blood Transfusion, Second Xiangya Hospital, Central South University, Changsha, Hunan Province, 410000 P, PR China

Ningjie Zhang

Key Laboratory of Sepsis Translational Medicine of Hunan, Central South University, Changsha, Hunan Province, 410000 P, PR China

You can also search for this author in PubMed   Google Scholar

Contributions

K.Z. supervised the whole project; K.Z. and L.L. designed the research; K.Z.L.L. and F.W. wrote the manuscript; L.L. and F.W. performed the experiments, analyzed the data and made the figures; X.X., M.M., G.K. and Y.Z. provided technical assistance; D.W. and W.L. assisted with the data analyses and discussions; and N.Z. assisted in the data interpretation and edited the manuscript.

Corresponding author

Correspondence to Kai Zhao .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Stat3 promotes nlrp3 inflammasome activation by mediating nlrp3 mito- chondrial translocation, supplementary table 1, rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Luo, L., Wang, F., Xu, X. et al. STAT3 promotes NLRP3 inflammasome activation by mediating NLRP3 mitochondrial translocation. Exp Mol Med (2024). https://doi.org/10.1038/s12276-024-01298-9

Download citation

Received : 15 December 2023

Revised : 10 May 2024

Accepted : 11 June 2024

Published : 02 September 2024

DOI : https://doi.org/10.1038/s12276-024-01298-9

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies