FacTool:
Factuality Detection in Generative AI

I-Chun Chern2, Steffi Chern2, Shiqi Chen3, Weizhe Yuan4, Kehua Feng1,
Chunting Zhou5, Junxian He6, Graham Neubig2, Pengfei Liu1,7*
1Shanghai Jiao Tong University 2Carnegie Mellon University 3City University of Hong Kong 4New York University 5Meta AI 6The Hong Kong University of Science and Technology 7Shanghai Artificial Intelligence Laboratory *Corresponding Author
Paper Code

Fact-checking the information generated by GPT-4 about Elon Musk, xAI, and his proposed cage fight with Mark Zuckerberg.

针对GPT-4所推荐的李白诗词进行事实性确认

GPT-4が生成したワールドベースボールクラシックの関連内容に対して事実確認を行う

FacTool

FacTool is an innovative, tool-augmented framework designed to detect factual errors in texts generated by Large Language Models (LLMs) across various scenarios. FacTool supports factual error detection across four tasks: knowledge-based question answering, code generation, math problem solving, and scientific literature review.

Figure 1
Figure 2

Abstract

The emergence of generative pre-trained models has facilitated the synthesis of high-quality text, but it has also posed challenges in identifying factual errors in the generated text.

In particular:
(1) A wider range of tasks now face an increasing risk of containing factual errors when handled by generative models.
(2) Generated texts tend to be lengthy and lack a clearly defined granularity for individual facts.
(3) There is a scarcity of explicit evidence available during the process of fact checking.

With the above challenges in mind, we propose FacTool, a task and domain agnostic framework for detecting factual errors of texts generated by large language models (e.g., ChatGPT).

FacTool now supports factual error detection of four different tasks: knowledge-based QA, code generation, math problem solving, and scientific literature review.

Code and ChatGPT plugin configs can be found here.

Knowledge-based QA

FacTool powered by GPT-4 fact-checking Knowledge-based QA responses on Barbenheimer generated by GPT-4.

Code Generation

FacTool powered by GPT-4 fact-checking code generated by GPT-4.

Math Problem Solving

FacTool powered by GPT-4 fact-checking math solutions generated by GPT-4.

Scientific Literature Review

FacTool powered by GPT-4 fact-checking scientific literature review generated by GPT-4.

BibTeX

@article{chern2023factool,
  title={FacTool: Factuality Detection in Generative AI--A Tool Augmented Framework for Multi-Task and Multi-Domain Scenarios},
  author={Chern, I-Chun and Chern, Steffi and Chen, Shiqi and Yuan, Weizhe and Feng, Kehua and Zhou, Chunting and He, Junxian and Neubig, Graham and Liu, Pengfei and others},
  journal={arXiv preprint arXiv:2307.13528},
  year={2023}
}