Let Me Check on You: Worker Well-Being Under AI and Human Health and Safety Monitoring

Artificial intelligence (AI) may improve workplace safety but could also reduce the non-pecuniary benefits of work. While research on workplace AI has focused mainly on productivity and employment, this paper presents causal evidence on how individuals evaluate hypothetical AI-based occupational health and safety systems. In a preregistered vignette experiment with Dutch adults ( N  = 2,172), respondents evaluated otherwise identical jobs with human supervision, AI-only monitoring, or hybrid AI-human supervision. Relative to human supervision, both AI-based systems reduced perceived job satisfaction, work meaningfulness, and social usefulness. Combining AI with human supervision did not attenuate these effects. Respondents also viewed both AI-based systems as less respectful of workers’ privacy and dignity, despite perceiving little difference in safety effectiveness. Fair-wage estimates provide no consistent evidence of a compensating wage differential for the perceived well-being costs of AI supervision. The welfare effects of workplace AI therefore extend beyond productivity, employment, and safety to include non-pecuniary dimensions of work well-being.

JEL Classification : I39, J01, J28, O33.