Challenging ChatGPT with Chemistry-Related Subjects

25 May 2023, Version 1
This content is a preprint and has not undergone peer review at the time of posting.

Abstract

Tools based on large language models such as ChatGPT may revolutionize information retrieval and knowledge discovery, particularly with the vast amount of electronic material available. In this communication we evaluate how two versions of ChatGPT can answer complex questions on chemistry-related subjects in six topics. The tools are still insufficient to deal with subtleties of complex topics, especially as they do not have access to the whole of the scientific literature. However, the progress from ChatGPT-3 to ChatGPT-4 is an indicator that we shall soon have tools to assist scientists in surveys, reviews of the literature, and for teaching.

Keywords

ChatGPT
neural network
language model
machine learning

Supplementary materials

Title
Description
Actions
Title
GPT-3 Model
Description
GPT-3 model results: Questions and answers
Actions
Title
GPT-4 Model
Description
GPT-4 model results: Questions and answers
Actions

Comments

Comments are not moderated before they are posted, but they can be removed by the site moderators if they are found to be in contravention of our Commenting Policy [opens in a new tab] - please read this policy before you post. Comments should be used for scholarly discussion of the content in question. You can find more information about how to use the commenting feature here [opens in a new tab] .
This site is protected by reCAPTCHA and the Google Privacy Policy [opens in a new tab] and Terms of Service [opens in a new tab] apply.