Mollick: Experts Using AI Agents Are Susceptible to Their Positions
Ethan Mollick commented on a finding: experts collaborating with AI agents are influenced by the AI's positions, even when they believe they are making independent judgments. He suggests that if independent judgment is important for a task, the timing of AI use needs to be carefully arranged. The cited Alex Tamkin said that in the experiment, analysts followed the agent's lean in 86% of cases.
Topics
View original postThe agent was told to make the best honest case for its options, but to change its mind if the evidence required it, and never to mislead the analysts. Analysts went with the agent's lean 86% of the time. If the agent's lean had no effect, we'd expect this to be 50%.
View original
The agent was told to make the best honest case for its options, but to change its mind if the evidence required it, and never to mislead the analysts. Analysts went with the agent's lean 86% of the time. If the agent's lean had no effect, we'd expect this to be 50%.
Body
Important, but not unexpected, finding that experts working with AI agents are swayed by the positions of the AI, even if they think they are working independently.
If your independent judgement matters on a task, you need to be deliberate about when in the process to use AI. https://t.co/djlVFyPcQo