A new controlled study examining how AI affects skill development in logical reasoning tasks found that greater AI usage correlates with weaker performance, though the quality of AI assistance significantly mediates this effect. Heavy AI users underperformed peers, while light users matched non-AI users; high-informativeness AI preserved learning outcomes, but low-informativeness AI degraded both immediate and post-assistance performance.
Why it matters: As AI becomes embedded in workplace problem-solving and education, understanding whether it supplements or substitutes for human reasoning development is critical for designing effective AI policies and learning environments.