GPTEval: A Survey on Assessments of ChatGPT and GPT-4

TL;DR

This survey comprehensively reviews existing assessments of ChatGPT and GPT-4 across various domains, highlighting their capabilities, limitations, and evaluation methods, and offers recommendations for future research in large language model evaluation.

Contribution

It provides a thorough synthesis of prior assessment studies of ChatGPT and GPT-4, analyzing their performance, evaluation techniques, and ethical considerations, which was lacking in previous literature.

Findings

ChatGPT and GPT-4 show strong language and reasoning abilities.

Current evaluation methods have limitations and need improvement.

Recommendations are provided for future assessment research.

Abstract

The emergence of ChatGPT has generated much speculation in the press about its potential to disrupt social and economic systems. Its astonishing language ability has aroused strong curiosity among scholars about its performance in different domains. There have been many studies evaluating the ability of ChatGPT and GPT-4 in different tasks and disciplines. However, a comprehensive review summarizing the collective assessment findings is lacking. The objective of this survey is to thoroughly analyze prior assessments of ChatGPT and GPT-4, focusing on its language and reasoning abilities, scientific knowledge, and ethical considerations. Furthermore, an examination of the existing evaluation methods is conducted, offering several recommendations for future research in evaluating large language models.