User-Centered AI Support for Mentorship: Perspectives From Autistic STEM Students in Higher Education

Autistic college students often face barriers to academic and career success, including lack of support during postsecondary education, leading to reduced graduation rates and difficulty accessing competitive employment. This is particularly relevant for science, technology, engineering, and mathematics (STEM) fields as autistic students participate in STEM at proportionally high rates. This exploratory study examined the results of a survey of 202 autistic undergraduate STEM students to understand the acceptability of, needs, and concerns regarding artificial intelligence (AI) use in STEM mentorship programs. Participants reported strong interest in AI integration into STEM mentorship and highlighted the importance of multimodal interaction, individualization, and accessibility. Participants also reported concerns, including issues of privacy and lack of personalization. Findings suggest autistic STEM students find AI tools to be a desired addition to traditional human-centered mentorship. However, caution and improvements are necessary to mitigate risk and better understand effective integration of AI into STEM mentorship.