The Information Machine

Anthropic's new study says frontier AI needs input from scholars, philosophers, clergy, and civic thinkers because model…

Rohan Paul Twitter · Rohan Paul (@rohanpaul_ai) · 2026-05-20

Anthropic published a study arguing that frontier AI development requires input from scholars, philosophers, clergy, and civic thinkers because Claude's trained behavior reflects questions of character and values beyond text prediction.

Open original ↗

Appears in

Extraction

Topics: ai-alignmentai-governanceanthropicai-valuesfrontier-ai

Claims

  • Anthropic argues that Claude's behavior is shaped by character and values, not only by text prediction objectives.
  • Later stages of training push frontier models toward value-laden behavior that goes beyond predicting the next token.
  • Anthropic believes scholars, philosophers, clergy, and civic thinkers should have meaningful input into frontier AI development.
  • Model behavior has become a question of character, requiring broader humanistic and societal input.

Key quotes

model behavior is becoming a question of character, not just code
Claude is not only trained to predict text, because later training pushes it