Lee, Sarkar, Tankelevitch et al. (2025) · CHI ’25, ACM · April 2025
AI & Critical Thinking
Survey evidence on when knowledge workers think critically while using generative AI.
4 in 10
AI-assisted tasks in which knowledge workers report engaging in no critical thinking
Survey of 319 knowledge workers describing 936 real AI-assisted work tasks; all measures are self-reported.
Perceived effort in critical-thinking activities falls for most task types; for comprehension tasks, 79% of examples were rated less effortful with AI.
Higher confidence in the AI is associated with less critical thinking; higher confidence in one’s own skills with more. The evidence is correlational.
Critical effort shifts rather than disappears: from producing content to verifying, integrating, and overseeing AI output.
Why it matters
The question for organizations is not whether AI removes thinking but where it relocates it: someone must still verify the output, and trust in the tool decides whether anyone does.
Peer-reviewed conference paper based on self-reported behavior. Source: Lee et al. (2025), The Impact of Generative AI on Critical Thinking. CHI ’25, ACM.