Can artificial intelligence produce a convincing accounting research article?

Abstract

PURPOSE :

This study aims to establish whether accounting research articles can be potentially generated by artificial intelligence. If artificial intelligence can produce quality work, the integrity of academic research may be compromised. DESIGN/METHODOLOGY/APPROACH :

ChatGPT was used to create a paper on a meta-analysis of the relationship between sustainability reporting and value relevance. After the paper was generated, references had to be added by hand based on the citations created by ChatGPT. The paper was then presented as-is for review. FINDINGS :

ChatGPT was able to create a relatively good-quality research paper that received two major revisions from independent specialists in the field of accounting and finance. Even though there is uncertainty regarding the appropriateness of all the references and the results cannot be confirmed, there is a risk that a reviewer may find the paper publishable because reviewers are not compelled to check references and the accuracy of results if proper methods were used that appear to be sufficient at face value. ORIGINALITY/VALUE :

Artificial intelligence for academic writing is still relatively new, and there is still significant uncertainty as to the impact it may have on scholarly research. This is especially problematic because artificial intelligence applications improve by the second.