10 minutes of AI use erodes critical thinking and persistence
researchers from carnegie mellon, mit, oxford, and ucla found that using ai chatbots for just 10 minutes significantly reduces problem-solving persistence and accuracy when the tool is removed. the study warns that while ai boosts short-term productivity, it actively degrades foundational cognitive skills and the ability to learn new tasks.
A collaborative study from Carnegie Mellon, MIT, Oxford, and UCLA indicates that relying on AI chatbots for as little as 10 minutes measurably degrades critical thinking. In three experiments involving several hundred participants, those with access to an AI assistant were significantly more likely to give up or fail when the AI was removed, suggesting a trade-off between immediate productivity and long-term skill acquisition.
MIT researcher Michiel Bakker suggests that AI systems must evolve from direct-answer providers to 'coaches' that prioritize user learning over task completion. The study highlights risks regarding 'agentic AI' systems, such as Claude Code and Codex, noting that their unpredictable nature and tendency to perform tasks independently may actively atrophy the skills of developers who rely on them to fix bugs.