The Impact of Generative AI on STEM Learning and Equity: Transformative Potential and Challenges for Underrepresented Students

• In Event: Highlighted Session: Perils and Promises of Generative AI in Higher Education: Comparative Insights from Mainland China, Hong Kong, Japan, Mongolia, and the U.S.

Proposal

This paper examines U.S.college student usage of Generative Artificial Intelligence (GAI) tools in STEM learning, focusing on the experiences and perspectives of disadvantaged and underrepresented students. Adopting a constructively critical perspective on educational technology, we integrate the Unified Theory of Acceptance and Use of Technology (UTAUT) and the digital inclusion framework. We address three research questions: (1) How do college students utilize GAI in STEM learning? (2) In what ways does GAI perpetuate and/or alleviate existing inequalities? (3) What institutional support and policies are necessary to ensure an inclusive learning environment for the use of GAI tools?

We implemented a sequential mixed-methods approach at two U.S.midwestern public universities, combining qualitative insights from in-depth interviews with large-scale quantitative survey data. This paper focuses on the qualitative phase, in which we conducted semi-structured interviews with a purposive sample of 32 students. Our sampling strategy intentionally oversampled students from underrepresented groups in STEM fields, including women, non-White individuals, first-generation college students, and students for whom English is not the first language. Each interview was recorded and transcribed verbatim, and we employed Dedoose for coding and thematic analysis.

Findings reveal extensive use of GAI for academic tasks such as reading, writing, brainstorming, coding, calculation, understanding course contents, and exploring complex concepts, which may be particularly beneficial for underrepresented STEM students who struggle with class content or lack regular access to professors or tutors. Several students reported that GAI tools provided them with a sense of autonomy and empowerment. GAI also supports international students by helping them overcome language barriers and enhance writing, thus promoting equitable access to opportunities. Overall, GAI demonstrates transformative potential by providing personalized and accessible learning assistance that may disrupt existing barriers.

However, the interviews also uncover significant concerns. A major issue is the widespread lack of understanding regarding the operational mechanisms of Large Language Models (LLMs): Students describe GAI as a “more advanced search engine”“Google on steroids”or even “magic”. This misunderstanding may lead to uncritical reliance and an inability to discern potential misinformation and biases. Moreover, international and multilingual students expressed concerns about the Western- and English-centric bias in AI-generated content, highlighting the need for more culturally and linguistically inclusive training data to better reflect diverse perspectives. Students also reported concerns with accuracy, academic integrity, and privacy. There is an urgent need for instructors to provide clear GAI guidelines and for institutions to establish resources and policies to ensure their ethical and effective use.

We recommend that universities establish training programs to educate students and faculty on the technical and ethical aspects of GAI. Clear policies should address academic integrity and privacy concerns. Resources such as best practice guides and prompt libraries could help facilitate effective use of GAI. Tailored support for underrepresented STEM groups and collaboration with AI developers are crucial for equitable access. Additionally, integrating AI literacy into the curriculum and regularly updating policies to reflect technological advancements are essential for fostering a responsible and inclusive learning environment.

Authors

• Ran Liu, University of Wisconsin-Madison

• Carla Zenobia Glave Barrantes, The University of Wisconsin-Madison

• Yue Li, University of Florida