What is Self-critique?
Self-critique is an agent-design concept in AI systems. It is part of the vocabulary teams use when they compare models, design prompts or review an AI system.
Definition
Self-critique is an agent-design concept in AI systems. It is part of the vocabulary teams use when they compare models, design prompts or review an AI system. The term is used in agents discussions, where precision matters: two teams can use the same word to mean slightly different things. In this glossary, self-critique is treated as a practical concept rather than a formal specification.
Why it matters
Agent terms describe how a model plans, calls tools, observes results and decides what to do next. They matter because agent reliability depends as much on control flow and guardrails as on raw model quality.
Example
Suppose you are comparing GPT-NeoX Instruct with another model for a production task. Self-critique is one of the factors behind the numbers you see: GPT-NeoX Instruct lists a 16,384-token context window, costs $1.52 per million input tokens and scores 75.4 on HellaSwag. Understanding the term helps you decide which of those differences actually matter for your workload.
How to evaluate it
- Check whether self-critique is defined the same way in the documentation you are reading.
- Ask which measurement or trade-off the term is describing before comparing numbers.
- Separate vendor claims from independently observed behaviour on your own data.
- Write down the definition your team uses so reviews and evaluations stay consistent.
Related terms
Related guides
FAQ
What does Self-critique mean?
Self-critique is an agent-design concept in AI systems. It is part of the vocabulary teams use when they compare models, design prompts or review an AI system.
Why does Self-critique matter?
Agent terms describe how a model plans, calls tools, observes results and decides what to do next. They matter because agent reliability depends as much on control flow and guardrails as on raw model quality.
How is Self-critique used in practice?
Suppose you are comparing GPT-NeoX Instruct with another model for a production task. Self-critique is one of the factors behind the numbers you see: GPT-NeoX Instruct lists a 16,384-token context window, costs $1.52 per million input tokens and scores 75.4 on HellaSwag. Understanding the term helps you decide which of those differences actually matter for your workload.