AI-HR: An Approach to Improve Performance of Large Language Model in the Pre-screening Process

Abstract

Listen

Translate article

This paper introduces the development of an HR-assisting website designed to streamline the pre-screening process using large language models. We improved the GPT APIs performance through Chain-of-Thought prompting and tailored evaluation rubrics to provide more detailed and accurate assessments of job candidates. We also conducted some experiments to investigate some ethical concerns, such as bias in the recruitment process. Based on the test results, we tried to formulate a strategic approach using GPT-4o mini to enhance fairness and inclusivity in the evaluation process.