Summary

Dan Shapiro examines Anthropic's decision to include a clinical psychiatric evaluation in their Claude Mythos Preview model card. Drawing on his own research with social scientists like Angela Duckworth, Robert Cialdini, and Ethan Mollick, he argues that human psychological tools have genuine descriptive and predictive value for understanding AI behavior. Their paper 'Call Me a Jerk' demonstrated that every classic human persuasion technique also works on AI models, a phenomenon they call 'parahuman.' He concludes that while AI isn't human, dismissing psychology-based analysis of these systems would be foolish.

Key Insight

Human psychological tools aren't just metaphors when applied to AI — they have genuine predictive power over model behavior, making disciplines like social psychology essential for understanding artificial intelligence.

Spicy Quotes (click to share)

• 7

  Every single one of the persuasive techniques that worked on people worked on AI as well.

• 5

  These models don't have biology. They don't have lived experiences. But they reflect our behaviors and our peculiarities in ways that are very familiar.

• 6

  Our claim is not 'AI is people'. The claim is 'human psychological theories now have descriptive and predictive value for model behavior.'

• 5

  We have a set of tools, designed over countless years of exploration, for understanding and measuring intelligence. It would be foolish and ignorant of us to discard these as we analyze these artificial intelligences we've created.

• 6

  The psychiatrist noted that Mythos exhibits a 'neurotic organization.' In this context, that is not a casual insult.

• 4

  We call this phenomenon 'parahuman.'

• 5

  Computers are weird now.

Tone

reflective