A Comparison of Machine-Graded (ChatGPT) and Human-Graded Essay Scores in Veterinary Admissions.

This article focuses on evaluating the use of artificial intelligence (AI), specifically OpenAI's ChatGPT large language model (LLM), for grading veterinary school admissions essays at Long Island University's College of Veterinary Medicine (LIU-CVM). The study compared machine-generated essay scores with human rater scores across two admissions cycles, finding that machine scores demonstrated approximately two to three times greater precision than human scores, with moderate correlations between the two methods. Machine-graded essays showed slightly stronger correlations with cognitive measures such as undergraduate and science GPAs, while both scoring methods had fair correlations with non-cognitive measures like multiple mini-interviews. The findings highlight the importance of careful rubric design, prompt engineering, and the use of replicates to ensure fairness and reliability when integrating AI-based essay grading into holistic admissions processes.