At a glance
- Source
- Lee, Sarkar, Tankelevitch, Drosos, Rintel, Banks and Wilson (2025). The Impact of Generative AI on Critical Thinking: Self-Reported Reductions in Cognitive Effort and Confidence Effects From a Survey of Knowledge Workers. Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems, Article 1121. doi:10.1145/3706598.3713778
- How it was done
- Knowledge workers answered questions once, each describing real examples of using AI at work, and said how much critical thinking each task took. Survey. People answered questions once. It shows what goes together, not what causes what.
- Who took part
- 319 knowledge workers, who gave 936 first-hand examples of using generative AI in their work.
- Where
- Knowledge work in many jobs; a survey by researchers at Microsoft Research.
What it found
- Trust in the tool. The more confident a worker was in generative AI on a task, the less critical thinking the worker reported on that task.
- Trust in oneself. The more confident a worker was in their own ability to do the task, the more critical thinking the worker reported.
- What the thinking became. With AI, workers described less effort on producing the work. They described more on checking information, combining AI answers into the work and overseeing the task as a whole.
When the result holds
Checking AI output is the job of the person who receives it. In this survey, that check weakened as trust in the tool grew. Trust in a tool tends to grow with use.
Limits
- Workers reported their own thinking. Nobody measured how well they checked.
- The survey asked once. It shows what goes together, not what causes what.
- The authors work for Microsoft, which sells AI assistants for office work. The finding runs against the company’s commercial interest.
Practices that use this study
This study supports:
This study limits or contradicts: