AI at Work: Working Arrangements and Human–AI Collaboration

Abstract

This exploratory study examines how AI adoption and AI-related attitudes vary across private-office, open-plan, hybrid, and remote working arrangements among knowledge workers. Drawing on a cross-sectional online survey of knowledge workers (N=60), the study examined associations between four working arrangements and 18 AI-related Likert items using item-level contingency tables, Pearson’s chi-square statistics, and Cramér’s V. Because sparse expected frequencies were common, conditional Monte Carlo estimates were used as the primary inferential values. Six items showed nominal cross-arrangement associations according to Monte Carlo inference (p<.05), with three medium and three large observed association magnitudes. Three preliminary patterns emerged. First, hybrid and remote respondents reported regular AI use more often than private-office and open-plan respondents, and perceived task efficiency showed a similar descriptive distribution. Second, instrumental framing of AI as a task-support tool was widely endorsed across arrangements, whereas assistant-like and collaborative framings varied across groups. Third, the strongest observed associations were organisational: employer provision or permission to use AI tools and reported freedom over AI use both varied across working arrangements. The pattern suggests that organisational conditions, including AI provision and autonomy of use, may help explain higher reported AI use in hybrid and remote arrangements. However, the cross-sectional design cannot distinguish this interpretation from alternatives such as organisational digital maturity, occupational differences, or worker selection. Keywords: hybrid work; remote work; human–AI collaboration; workplace layout; knowledge work