Simplifying Peer Review for Accessibility: A Case Study on GPT 4's Performance in Finance Using Cognitive-Informed Prompts

This study investigates how GPT-4 performs in simplifying peer reviewer comments in the field of academic finance, particularly for cognitively diverse users such as those with dyslexia or working memory challenges. Using ten domain-specific reviewer comments and two types of prompts—one general and one tailored for cognitive accessibility—the paper evaluates 40 GPT-4 outputs for accuracy, consistency, and conceptual fidelity. The results reveal frequent misrepresentations of key technical terms, oversimplifications that distort meaning, and inconsistency across runs, regardless of prompt type. The findings highlight critical limitations in using large language models for accessibility-focused simplification in specialized academic domains and call for more domain-sensitive AI development.