Loading patterns…
Constitutional AI(CAI)
Using an explicit set of principles to guide self-critique, revision, and AI-generated preference feedback during training
In 30 seconds
- What
- Model learns to critique and revise its own outputs against explicit principles, then ranks responses using principle-aligned feedback without human labeling.
- When to use
- Training systems where you have clear behavioral principles but limited human feedback capacity, and need interpretable alignment during learning.
- Watch out
- Principles that conflict or are vague produce inconsistent training signals; model may learn to game principle-checking rather than internalize values.
Ask the AI expert about this pattern
Opens the assistant with your question prefilled. You review it before sending.
Constitutional AI: Overview
Using an explicit set of principles to guide self-critique, revision, and AI-generated preference feedback during training
- Explicit constitutional principles
- Self-critique and revision data
- Principle-conditioned preference judgments
- Supervised and reinforcement learning stages
- Conflict and ambiguity testing
- Human governance of the constitution
The Agent Architect
One pattern, one tradeoff, one production failure story. A short weekly briefing for people building agentic systems.
Weekly email, one-click unsubscribe. We only use your address to send the briefing.
References
The papers, specifications, and repositories this pattern is based on.
- Constitutional AI: Harmlessness from AI Feedback (2022)arXiv:2212.08073
From the engineer behind this catalog
Get your agent architecture reviewed
This page documents one pattern. Your system runs dozens, and most failures live in how they fit together. Have the whole design reviewed against the 288 patterns in this catalog: architecture, reliability, evaluation and cost, every finding mapped to the pattern that fixes it.
€750 instead of €1,500, one week, written report and walkthrough call, until 30 September