Algorithms and the future of work

Abstract

An algorithm refers to a series of stepwise instructions used by a machine to perform a mathematical operation. In 1955, the term artificial intelligence (AI) was coined to indicate that a machine could be programmed to duplicate human intelligence. Even though that goal has not yet been reached, the use of sophisticated machine learning algorithms has moved us closer to that goal. While algorithm‐enabled systems and devices will bring many benefits to occupational safety and health, this Commentary focuses on new sources of worker risk that algorithms present in the use of worker management systems, advanced sensor technologies, and robotic devices. A new “digital Taylorism” may erode worker autonomy, and lead to work intensification and psychosocial stress. The presence of large amounts of information on workers within algorithmic‐enabled systems presents security and privacy risks. Reliance on indiscriminate data mining may reproduce forms of discrimination and lead to inequalities in hiring, retention, and termination. Workers interfacing with robots may face work intensification and job displacement, while injury in the course of employment by a robotic device is also possible. Algorithm governance strategies are discussed such as risk management practices, national and international laws and regulations, and emerging legal accountability proposals. Determining if an algorithm is safe for workplace use is rapidly becoming a challenge for manufacturers, programmers, employers, workers, and occupational safety and health practitioners. To achieve the benefits that algorithm‐enabled systems and devices promise in the future of work, now is the time to study how to effectively manage their risks.

Artificial intelligence and the work–health interface: A research agenda for a technologically transforming world of work

“…29 However, AI can also contribute to workplace hazards for human workers such as the intensification of work and performance pressures (e.g., increasing pace of work to keep up with productive AI), loss of job autonomy (e.g., AI takes control over the performance of job tasks) and the psychosocial stressors arising from the fear of being made obsolete as AI fully automates a job. 8,30 Also, the disproportionate use of AI across different occupations will have a variable impact on the demand for workers and may contribute to inequities in the labor market. [31][32][33][34][35][36] For instance, some estimate that occupations least likely to be exposed to AI are characterized by unpredictable and highly physical job tasks (e.g., nursing aide, janitor, food service worker) or require strategic leadership responsibilities (e.g., chief executive officer).…”

“…22 Some literature highlights the use of AI to rapidly predict potential workplace risks or safety incidents, diagnose the root cause of an OHS incident, and support evidence-based OHS decision-making. 30 AI is also being used by workplaces to identify changes to worker physical health and mental health that provide real-time feedback and support, and promote worker health-enhancing behaviors. On the other hand, the large amounts of data required for the development, training, and refinement of AI tools to promote health and safety can be a source of stress for workers who may reject constant surveillance or express concerns over the collection and storage of personal information.…”

“…The design of algorithms that underpin AI function has the potential to produce discriminatory outcomes if they do not account for the diversity of worker experiences and backgrounds. 30,38 Numerous studies of algorithmic bias highlight how gaps in their design can reproduce and reinforce social and health inequities with significant implications for the most disadvantaged groups of workers. 39,40 Similarly, datasets used to train AI that lack representation or diversity T A B L E 1 Summary of potential health impacts of AI on worker health, safety, well-being, and equity.…”

“…AI within the workplace can be used to automate physically and psychologically demanding job tasks 29 . However, AI can also contribute to workplace hazards for human workers such as the intensification of work and performance pressures (e.g., increasing pace of work to keep up with productive AI), loss of job autonomy (e.g., AI takes control over the performance of job tasks) and the psychosocial stressors arising from the fear of being made obsolete as AI fully automates a job 8,30 . Also, the disproportionate use of AI across different occupations will have a variable impact on the demand for workers and may contribute to inequities in the labor market 31–36 .…”

“…AI can be used by workplaces as a tool to address the hazards to which workers are exposed and promote health 22 . Some literature highlights the use of AI to rapidly predict potential workplace risks or safety incidents, diagnose the root cause of an OHS incident, and support evidence‐based OHS decision‐making 30 . AI is also being used by workplaces to identify changes to worker physical health and mental health that provide real‐time feedback and support, and promote worker health‐enhancing behaviors.…”