Globally, employers across diverse industries are turning to artificial intelligence (AI) to innovate workplace processes and automate job tasks (1). Many of these AI applications are becoming increasingly independent from human operators and play prominent roles in making decisions and acting within the workplace. In this way, AI is contributing to shifts in work environments that may have significant implications for worker health, safety, and well-being (2). Governments are seeking to stimulate the commercialization of AI technologies and development of AI infrastructure to boost economic productivity (3). While there is an acknowledgment of the societal importance of creating safe, responsible, and pro-worker AI to mitigate risks that the technology may pose to workers and amplify its benefits, the health of workers has largely been absent from governmental agendas (4, 5). While there has been a growing focus by occupational health and safety (OHS) researchers on the impact of advanced digital technologies (6) and algorithmic management (7) on the health and safety risks faced by workers, studies of AI-specific transformations to the work environment and the health and safety of workers remain limited (8). Consequently, there is a paucity of evidence for the design of specific actionable strategies to protect and promote the health and safety of workers who are exposed to AI and AI-enabled systems in their working lives. Understanding how these technologies are transforming the relationship between work, health, and safety is an urgent scholarly priority for OHS researchers to address.

Artificial intelligence at work
AI is an umbrella term for computing systems that perform tasks that have traditionally required human intelligence including recognizing patterns, making predictions, or generating decisions (9). Most modern AI systems rely on algorithms that learn patterns from large training datasets rather than following explicit, hand-coded rules. Machine learning, a core subfield, encompasses methods for this data-driven learning, with deep learning, a further subfield of machine learning that uses multi-layered neural networks, handling more complex patterns. Other subfields apply these techniques to specific data types, such as natural language processing (text) and computer vision (images and video).
Examples of AI adoption to perform specific job tasks exist in virtually all industries including those which have traditionally been open to adopting new technologies (eg, healthcare and manufacturing) and those less prone to technological change (eg, arts and culture) (10, 11). While in the early stages, AI is also being adapted to OHS activities within workplaces, such as in tools that monitor safety compliance in high-risk settings or promote healthy behaviors among workers (8). These examples have yet to show a mass disruption to the availability of jobs but can result in a re-organization of work between humans and intelligent machines that is fundamentally changing the way workers perform their jobs as well as their employment and work conditions (12). While research is scant, in table 1, we have synthesized available literature to describe how AI systems may impact different dimensions of the work environment such as work arrangements, and physical and psychosocial conditions.

OHS AI research challenges through the lens of the Partnership on AI and Quality of Work (PAIQ)
PAIQ is a seven-year project based at the Institute for Work & Health in Toronto, Canada that brings academic researchers together with multi-sectoral stakeholders to unpack how AI adoption impacts work and employment conditions and worker health and well-being (13). Drawing from our experience in PAIQ, we present three principal challenges that OHS researchers may face when engaging in research on AI in the workplace.

Challenge 1: Rapid speed of AI advancement. The applications and abilities of AI have and continue to change very rapidly with many frontier models surpassing human-level performance (14). Attributed to scalable architecture, improved computing power and infrastructure, and significant financial investment, AI has increased in its processing speed, ability to learn from diverse datasets, and capacity to perform tasks with increasing autonomy. AI advancements will contribute to an emerging number of applications within the workplace and thus ongoing changes to different dimensions of the work environment.

Many occupational hazards (eg, chemical, ergonomic, physical, and psychosocial) can have a fixed impact on work environments. In contrast, AI’s ability to learn and adapt may result in the technology constantly shifting its impact on the health and safety profile of an organization. These shifts may be exacerbated by AI advancement that can result in the technology reasoning and acting with a decreasing amount of human input. Keeping up with the pace of AI technological change is challenging, especially within the constraints of traditional time-intensive academic research protocols. There is a risk that the production of rigorous evidence will be obsolete by the time the findings are published.

To match the speed of AI advancement with the ability of OHS scholars to study its impacts, dynamic data collection approaches are needed, such as yearly scans of evidence sources to identify emerging threats and opportunities of AI in the workplace. As an extension of this strategy, PAIQ’s use of a living systematic review methodology allows for the tracking of peer-reviewed research on new forms of AI and their impact on work environments and different worker health outcomes (15).

Challenge 2: AI invisibility and difficulties with primary data collection. AI is commonly embedded in existing digital technology within a workplace such as automated vehicles, mobile phones, robots, or the Internet-of-Things devices, and its impact on working experiences is not always visible to workers encountering the technology (2). In the same vein, employers choosing to adopt AI systems may have less clarity on how the technology is functioning and affecting workers. Despite its potential invisibility, AI may be actively dictating work task performance and/or pace of work, monitoring workers’ performance, and making decisions. Workers and employers participating in OHS research may struggle to identify how AI is affecting different dimensions of work and employment conditions, creating challenges for primary data collection (eg, surveys or qualitative interviews). Therefore, OHS research can benefit from diverse disciplinary lenses—that is, from both computer and data sciences as well as the social sciences and health sciences—to better understand how AI design elements as well as workplace characteristics (eg, social structures, institutions, and power relations) intersect to contribute to AI’s impact on work environments and worker health.

Challenge 3: AI upskilling for OHS researchers and practitioners. Computer and data science are not typically part of the OHS researcher’s sphere of knowledge. OHS scholars may be unprepared to examine how algorithm design, data selection and training, and deployment decisions may play a specific role in changing work conditions and worker health. Lacking deeper AI knowledge creates challenges to proposing and pursuing research that can lead to the identification of AI hazards and understanding the mechanisms of harm. We also run the risk of developing applied solutions that do not address the salient challenges AI may pose to worker health within workplaces.

OHS researchers who gain greater familiarity with how different AI subtypes and systems behave, fail, and interact with humans on the job may be better equipped to study how the design, training, and application of AI can change different dimensions of the work environment and identify why and how the technology may positively or negatively impact worker health. The need for gaining expertise on AI may be particularly pronounced with the growing deployment of frontier systems that have the potential to more drastically change job performance and dimensions of the work environment.

Concluding remarks
The adoption of AI at work creates new challenges and opportunities for the health, safety, and well-being of workers (16). The OHS sector offers a valuable perspective to the emerging discourse on AI safety and pro-worker AI by drawing greater attention to the intersection between the work environment and worker health (17). Importantly, OHS expertise has the potential to offer practical solutions that can be used to safeguard workers from novel AI-related hazards that may contribute to injury or illness. For OHS researchers to fully engage in the research and provide expertise to the AI research landscape, there is a need to transform the ways in which research is conducted to deploy dynamic research approaches, account for the invisibility of AI, and build AI literacy. While taking these steps requires an investment, they will enable OHS researchers to play a more prominent role in making an AI future healthier and safer for workers.

Acknowledgments
The study was funded by a Social Sciences and Humanities Research Council of Canada (SSHRC) Partnership Grant (#895-2025-1000) supporting PAIQ (www.iwh.on.ca/projects/partnership-on-ai-and-quality-of-work). We would like to acknowledge contributions from our multidisciplinary research team and multi-sector partners who are part of PAIQ. Authors are employees of the Institute for Work & Health (IWH), which is supported through funding from the Ontario Ministry of Labour, Immigration, Training and Skills Development (MLITSD). The analyses, conclusions, opinions, and statements expressed herein are solely those of the authors and do not reflect those of the MLITSD; endorsement is not intended nor should it be inferred.

References
1. Mayer H, Yee L, Chui M, Roberts R. Superagency in the workplace: empowering people to unlock AI’s full potential [Internet]. 2025 [cited 2026 August 13]. Available from: https://www.mckinsey.com/capabilities/tech-and-ai/our-insights/superagency-in-the-workplace-empowering-people-to-unlock-ais-full-potential-at-work.
2. Jetha A, Bierbrier J, Smith P. The AI shift: reconceptualizing health, safety, and equity in the future of work. In: Vanroelen C, Seubert L, Bodin T, editors. Occupational Health in a New World of Work. Cham: Springer Nature Switzerland; 2026. p. 1–22.
3. Government of Canada. Pan-Canadian artificial intelligence strategy [Internet]. Ottawa, ON 2025 [cited 2026 August 13]. Available from: https://ised-isde.canada.ca/site/ised/en/pan-canadian-artificial-intelligence-strategy.
4. Acemoglu D, Autor D, Johnson S. Building pro-worker artificial intelligence. The Hamilton Project; 2026 [cited 2026 August 13]. Available from: https://www.brookings.edu/wp-content/uploads/2026/02/20260223_THP_ProWorkerAI_Paper.pdf.
5. European Commission. AI Act. European Commission [Internet]. 2025 [cited 2026 August 13]. Available from: https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai.
6. Håkansta C, Asp A, Palm K. Effects of work-related digital technology on occupational health in the public sector: A scoping review. Work 2025 Jun;81(2):2477–90. https://doi.org/10.1177/10519815251320274.
7. Bowdler M, Lahti H, Jelenko M, Buresti G, Valtonen T. Algorithmic management and psychosocial risks at work: an emerging occupational safety and health challenge. Scand J Work Environ Health 2026 Jan;52(1):1–5. https://doi.org/10.5271/sjweh.4270.
8. Jetha A, Bakhtari H, Irvin E, Biswas A, Smith MJ, Mustard C et al. Do occupational health and safety tools that utilize artificial intelligence have a measurable impact on worker injury or illness? Findings from a systematic review. Syst Rev 2025 Jul;14(1):146. https://doi.org/10.1186/s13643-025-02869-1.
9. Stryker C, Kavlakoglu E. What is artificial intelligence (AI) [Internet]. IBM Think; 2024 [cited 2026 August 13]. Available from: https://www.ibm.com/think/topics/artificial-intelligence.
10. Beyai J. AI generated art implications [Internet]. Cultural Policy Hub, OCAD University; 2023 [cited 2026 August 13]. Available from: https://culturalpolicyhub.ocadu.ca/news/ai-generated-art-implications.
11. Deloitte. The AI Dossier: 125+ AI use cases [Internet]. 2026 [cited 2026 August 13]. Available from: https://www.deloitte.com/us/en/what-we-do/capabilities/applied-artificial-intelligence/content/ai-use-cases.html.
12. Massenkoff M, McCrory P. Labor market impacts of AI: a new measure and early evidence [Internet]. Anthropic; 2026 [cited 2026 August 13]. Available from: https://www.anthropic.com/research/labor-market-impacts.
13. Institute for Work and Health. Partnership on AI and the quality of work (PAIQ) [Internet]. Institute for Work & Health; [cited 2026 August 13]. Available from: https://www.iwh.on.ca/projects/partnership-ai-quality-of-work.
14. Sajadieh S, Fattorini L, Perrault R, Gil Y, Parli V, Santarlasci L et al. 2 Technical performance [Internet]. The AI Index 2026 Annual Report. AI Index Steering Committee, Institute for Human-Centered AI, Stanford University, Stanford, CA; 2026 [cited 2026 August 13]. Available from: https://hai.stanford.edu/assets/files/ai_index_report_2026_chapter_2_technical.pdf.
15. Jetha A, Crouch M, Vold K, Peters SE, Vietas J, Sriharan A et al. Artificial intelligence in the workplace: a living systematic review protocol on worker safety, health, and well-being implications. Syst Rev 2025 Dec;14(1):255. https://doi.org/10.1186/s13643-025-03000-0.
16. Jetha A, Bakhtari H, Rosella LC, Gignac MA, Biswas A, Shahidi FV et al. Artificial intelligence and the work-health interface: A research agenda for a technologically transforming world of work. Am J Ind Med 2023 Oct;66(10):815–30. https://doi.org/10.1002/ajim.23517.
17. Probst T, Shoss M, Bankins S, Behrend T, Dragano N, Jetha A et al. Aligning artificial intelligence with worker health, well-being, and safety: an occupational health psychology agenda. Occup Health Sci 2026;10(2):15. https://doi.org/10.1007/s41542-026-00262-5.

Key terms AI; artificial intelligence; occupational health and safety; research