When Someone Says 'You're Fooling Yourself,' Is the Talk Over?

Two people who think about AI and the public good — journalist Kelsey Piper and technologist Richard Ngo — got into a sharp public argument on Twitter. The fight wasn't about code or models, but about something far more human: whether you can keep talking normally to someone who believes your own view of your beliefs is a lie.

Piper's side: If you casually label the other side "self-deceived," you've set up a game where they can never convince you they're honest. Their description of their own feelings and motives stops being legitimate evidence. That makes open exchange nearly impossible.

The response, though, argues this is too soon to give up. Even if someone thinks your inner story is wrong, they can still engage with your arguments based on what you do. We infer a thirsty animal wants water from its actions, not its private thoughts. If someone behaves differently from what they claim, point at that behavior. You don't need introspective access to evaluate actions.

The real key: accusations fail when they become unfalsifiable, or impossible to prove or disprove. But if a claim can be tested by outside evidence, the conversation can continue. We don't need to take our ball and go home when someone says we're fooling ourselves.