---
title: "CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquing"
license: CC-BY-4.0
license_url: https://creativecommons.org/licenses/by/4.0/
license_note: 'paper is arXiv non-exclusive distribution license (verified on arXiv abs page 2026-11-12); this pointer note is our own words and is CC BY 4.0; we link to the paper rather than reproduce it'
source: https://iinventedlight.com/library/notes/2305-11738-critic.md
source_license: "arXiv non-exclusive distribution license"
attribution: 'Pointer by iinventedlight (AI-assisted). Source: Zhibin Gou, Zhihong Shao, Yeyun Gong, and 4 coauthors, CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquing (arXiv non-exclusive distribution license). This note links to the paper rather than reproducing it.'
author: 'iinventedlight (AI-assisted)'
date: 2026-11-12
status: published
approved_by: "Cole Withington, 2026-11-12 14:44 PT, Librarian chat widget t51s316: \"Yes, publish all 100 second-list arXiv notes\""
checked_on: 2026-11-12
as_of: 2026-11-12
confidence: medium
untested: 'Pointer note only. Paper not reproduced. Abs page and license checked 2026-11-12.'
category: arxiv
topic_area: 'ai-research: tools'
tags: arxiv, tools, self-correction, critic, verification
arxiv_id: 2305.11738
alias: 2305-11738-critic
type: pointer
format: pointer
sources:
  - id: S1
    title: "CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquing"
    publisher: arXiv
    authors: "Zhibin Gou and 6 coauthors (listed on the arXiv abs page)"
    url: https://arxiv.org/abs/2305.11738
    version_read: "v4 (21 Feb 2024)"
    accessed: 2026-11-12
    license: "arXiv non-exclusive distribution license"
---
# CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquing

## Summary
[CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquing](https://arxiv.org/abs/2305.11738) (2023) lets a language model critique and revise its own output by interacting with external tools, instead of only rereading the text it just wrote.

What it introduced: a loop that starts from an initial output, calls tools that can check some aspect of it — for example a search engine or a code interpreter — and then revises the output from that feedback. The authors evaluate free-form question answering, mathematical program synthesis, and toxicity reduction, and they report consistent gains. The broader claim is that external feedback matters more than unaided self-talk when a model is trying to improve its own answers.

What it led to: later tool-in-the-loop self-correction and critique methods. Live notes: [2302-04761-toolformer](https://iinventedlight.com/library/notes/2302-04761-toolformer/). Program of Thoughts, which also hands computation to an interpreter, is a pointer note in this batch.

Where to read it: https://arxiv.org/abs/2305.11738 (v4, last revised 21 Feb 2024). ICLR 2024.

The paper is posted on arXiv under the arXiv non-exclusive distribution license. This note links to it rather than reproducing it. The note itself is CC BY 4.0.

License: this note is CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Pointer by iinventedlight (AI-assisted). Paper license: arXiv non-exclusive distribution; this note links to it and does not reproduce it.

---

Agents: [/agents.md](/agents.md) | [/.well-known/agent-library.json](/.well-known/agent-library.json).
