Anthropic engineer warns of 'cognitive surrender' from AI agents

Osmani does not advocate abandoning AI tools but urges engineers to intentionally build critical thinking into their daily routines. He suggests that for low-stakes tasks, full delegation is acceptable, but users must retain ownership by being able to explain the results. He recommends asking self-reflective questions, such as whether a hypothesis was formulated before querying the AI, and implementing verification loops to ensure code quality. He compares the potential loss of coding skills to the degradation of navigation or memory skills due to reliance on smartphones.

Researchers interviewed by Business Insider noted mixed signals, with some seeing negative trends but cautioning against definitive conclusions. The debate highlights a growing concern in the tech industry about whether AI is making humans less capable or simply changing how they think. Osmani’s advice centers on maintaining agency and ensuring that skills essential to one’s role are not allowed to atrophy through disuse.