Anthropic co-founder on the risk of people turning into zombies

Anthropic co-founder Jack Clarke gave a lecture at Oxford about the future of AI. He warned that over-reliance on AI advice erodes human agency. As an example, he cited teenagers who coordinate their actions entirely based on Claude prompts.

According to Clarke, constantly delegating decisions to AI leads to a loss of independent thinking skills. Humans risk becoming philosophical zombies who act correctly but don't make their own mistakes or learn from them.

To address this issue, Clarke suggests implementing limits on AI systems, similar to those on gaming consoles. Models should proactively suggest that users take breaks and discuss personal issues with real people, not algorithms.