Skip to content

AI Safety & Responsible AI

Constitutional AI

A way of teaching AI to follow written principles, partly by using AI to review and improve replies.

Example

An AI system critiques a response against specified principles before it is used in training.

Why people use it

It uses stated principles to guide how AI is taught to respond.

What you'll hear

“Which principles is the AI being taught to follow?”

What this means for you

Inspect both the principles and evidence of the AI system's actual behavior.

Can you control it?

Developer-only

The people building or running the AI choose this setup. An everyday user generally needs their help to change how this part works.

Common questions

Does a written constitution ensure the AI system follows every principle?
No. Training can improve behavior without guaranteeing following the rules.
Are the principles the same for every system?
No. Different builders can choose different principles and ways of applying them.
Does it remove the need for human judgment?
No. People still choose the principles and assess whether the resulting behavior is acceptable.

Related terms

Still have questions?

Up to 500 characters.

Ask LATHIC about AI. Relevant glossary entries may be included.

Your question, the glossary entries it matches, and a rotating pseudonymous identifier go to Microsoft Azure’s OpenAI service through Vercel AI Gateway to generate an answer. Zero retention and no training are required of the provider, and LATHIC does not save your question or answer. Privacy Notice