2025 ASEE Annual Conference & Exposition

The number of applicants to universities in the US is steadily increasing every year, with undergraduate applications rising by 41% in 2023-24 from 2019-20. The increasing number of applications has led to several universities adopting AI in their admissions processes for various tasks such as evaluating transcripts, extracting important information from essays, and scoring applications. Past implementations of AI for data-driven decision-making for selection processes such as student recruitment have often led to issues surrounding bias with the potential to have long-standing effects on diversity in STEM. With the minimal change in diversity in engineering classrooms at the undergraduate and graduate levels over the last decade and the recent removal of affirmative action for admissions, it is critical to evaluate the potential outcomes of using AI in admissions, especially towards class diversity. In this paper, we propose that while AI could be leveraged to increase the efficiency of the decision-making process for student admissions, it is imperative to continuously identify and tackle bias that emerges from AI-based systems for admissions. Based on previous work, we identify some key sources of bias in the context of admissions and the role of AI in predicting outcomes and assisting decision-making. We discuss potential opportunities to leverage AI in the admission pipeline and identify its potential risks to diversity. We also illustrate various methods that can be applied to detect and alleviate emergent biases to foster diversity among STEM graduates.