Barriers to AI Adoption: Image Concerns at Work
David Almog
TL;DR
This study addresses how image concerns shape AI adoption in the workplace by running a field experiment on Upwork with 450 US-based data annotators. A two-stage task pairs unaided judgments with AI recommendations and varies whether AI reliance is observable to an HR evaluator, enabling causal inference about signaling effects on behavior and performance. The findings show that making AI reliance observable reduces AI adoption by about $4.3$ percentage points and lowers accuracy by roughly $3.4\%$, with workers spending more time on initial decisions but not improving final accuracy. A novel incentive-compatible public-feedback mechanism reveals that workers fear signaling low confidence in their own judgment when AI use is visible, and information-based reassurance does not alleviate this effect; these image concerns persist even under controlled evaluation criteria, suggesting substantial practical barriers to AI–human collaboration. The results imply that improving AI productivity requires addressing the social meanings attached to AI use or redesigning workflows to reduce visibility, and the study contributes a new platform-based signaling method for studying signaling motives in digital labor markets.
Abstract
Concerns about how workers are perceived can deter effective collaboration with artificial intelligence (AI). In a field experiment on a large online labor market, I hired 450 U.S.-based remote workers to complete an image-categorization job assisted by AI recommendations. Workers were incentivized by the prospect of a contract extension based on an HR evaluator's feedback. I find that workers adopt AI recommendations at lower rates when their reliance on AI is visible to the evaluator, resulting in a measurable decline in task performance. The effects are present despite a conservative design in which workers know that the evaluator is explicitly instructed to assess expected accuracy on the same AI-assisted task. This reduction in AI reliance persists even when the evaluator is reassured about workers' strong performance history on the platform, underscoring how difficult these concerns are to alleviate. Leveraging the platform's public feedback feature, I introduce a novel incentive-compatible elicitation method showing that workers fear heavy reliance on AI signals a lack of confidence in their own judgment, a trait they view as essential when collaborating with AI.