Skip to content

Model Training & Adaptation

Actor-critic

An AI learning setup with one part choosing actions and another judging their likely future rewards.

Example

One part of an AI chooses a move while another estimates how much reward it could lead to.

Why people use it

It divides learning between choosing actions and judging how promising situations are.

What you'll hear

“One part chooses; another estimates how well things are going.”

What this means for you

Remember that both parts may be software; arrange human oversight separately.

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 person have to act as the critic?
No. The critic is usually another learned part of the AI that estimates future rewards.
Can poor reward estimates mislead the part choosing actions?
Yes. Inaccurate judgments can push the action chooser toward less useful behavior.
Do the two parts have to be separate physical systems?
No. They are roles within a learning setup and can share some basic components.

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