{"id":35695,"date":"2026-03-17T15:49:38","date_gmt":"2026-03-17T15:49:38","guid":{"rendered":"https:\/\/machinelearningplus.com\/uncategorized\/constitutional-ai-self-critique-python\/"},"modified":"2026-03-17T19:27:38","modified_gmt":"2026-03-17T19:27:38","slug":"constitutional-ai-self-critique-python","status":"publish","type":"post","link":"https:\/\/machinelearningplus.com\/gen-ai\/constitutional-ai-self-critique-python\/","title":{"rendered":"Constitutional AI: Build a Self-Critique Loop (Python)"},"content":{"rendered":"

Generate a response, critique it against principles, and revise \u2014 all with raw HTTP calls to an LLM API.<\/em><\/p>\n

\nInteractive Code Blocks<\/strong> \u2014 The Python code blocks in this article are runnable. Click the Run<\/strong> button to execute them right in your browser.\n<\/div>\n

Your chatbot just told a user how to pick a lock. You didn’t ask it to. The model has no filter \u2014 it answers whatever you throw at it. Now imagine a system that catches that before the response leaves the server. No human reviewer. The model critiques itself.<\/p>\n

That’s Constitutional AI. Anthropic introduced it in 2022 to align models without drowning in human feedback. The model generates, checks against principles, and rewrites. One model, three passes, safer output.<\/p>\n

Here’s how the pieces connect. You send a prompt. The model generates a raw response with no guardrails. You feed that response back with a principle like “Is this harmful?” The model critiques its own answer. Then you send the original plus the critique and ask for a revision. Each stage feeds the next: raw response flows into critique, critique flows into revision.<\/p>\n

We’ll build this loop from scratch with raw HTTP calls.<\/p>\n

What Is Constitutional AI?<\/h2>\n

Think of it this way. You write five rules on a card: no harm instructions, no fake facts, no bias, no privacy leaks, no manipulation. You hand the card to the model alongside its own response. “Did you break any of these?” The model checks each rule and flags violations.<\/p>\n

Constitutional AI (CAI) is this pattern turned into a pipeline. Anthropic published the original paper in December 2022. The model evaluates its own output against explicit, written principles \u2014 then rewrites to fix what it found.<\/p>\n

Standard alignment needs thousands of human preference labels. CAI replaces most of that labor with self-critique. The model still needs instruction-following ability. But the safety layer runs on the model’s own judgment.<\/p>\n

Key Insight:<\/strong> Constitutional AI replaces human safety reviewers with a structured self-critique loop: generate, critique against principles, then revise.<\/strong> The model polices itself.<\/div>\n

Prerequisites<\/h2>\n