Anthropic engineer warns AI agents risk 'cognitive surrender'
Addy Osmani said none of his daily work forces critical thinking anymore, and warned that unchecked reliance on AI agents builds cognitive debt.
Anthropic engineer Addy Osmani published comments on November 7, 2026 warning that heavy reliance on AI agents risks what he called "cognitive surrender" and the erosion of critical thinking skills, according to techradar.com.
Osmani said that none of his day-to-day work now forces critical thinking on him, because so much of it runs through agents. He warned that skills people stop exercising will fade, likely faster than they expect.
His central concern is cognitive debt: it builds when AI output is accepted without checks and users can no longer explain that output. Osmani argued that responsibility still rests with a human, regardless of how much AI helped.
He said some cognitive offloading to AI is acceptable as long as people keep ownership of the work. Among his suggestions were practising core skills, forming ideas before asking AI, paying attention to generated output and understanding how the AI reached its decisions.
Osmani added that some skills will inevitably be lost as AI improves, and that people should deliberately choose which ones to let go. The broader discussion of cognitive faculties degrading through underuse in the age of AI predates his remarks.
Quick answers
What did Addy Osmani mean by cognitive surrender?
Osmani used the term for heavy reliance on AI agents that removes critical thinking from daily work, and warned that unused skills fade faster than people expect.
What is cognitive debt in AI-assisted coding?
According to Osmani, it builds when AI output is accepted without checks and users can no longer explain that output.