Claude's constitution

Summary

Claude’s constitution describes how explicit principles can guide a language model’s behavior instead of relying on values implicitly determined through large-scale human feedback. The approach uses AI feedback in two training phases: the model first critiques and revises responses using the principles, then reinforcement learning selects the more harmless output using AI-generated evaluations. The post says Constitutional AI produced a Pareto improvement in tests, with the model responding more appropriately to adversarial inputs while remaining helpful and not evasive; harmlessness results came without human harmlessness data. The constitution draws on the UN Declaration of Human Rights, safety practices, other AI-lab principles, global platform guidelines, and attempts to include non-Western perspectives, while remaining an iterative, adjustable document.

Takeaways

1. 1

   Constitutional AI uses AI supervision rather than human harmlessness data to train responses to adversarial inputs.

2. 2

   The model’s principles are intended to make its behavior easier to specify, inspect, understand, and adjust.

3. 3

   The constitution combines sources including the UN Declaration of Human Rights, safety practices, other AI-lab principles, platform guidelines, and perspectives beyond Western cultures.