From learners to contributors: how an AI-infused STEM program shaped youth identity and initiated them to an AI-future

The expanding role of Artificial Intelligence (AI) across society has created new demands for K–12 education to support young people in developing foundational AI literacies. Although “AI education” remains in its infancy, an emerging field is taking shape that integrates technical learning, authentic applied experiences, and ethical considerations. This study reports findings from SRMPmachine, an out-of-school-time intervention that infused machine learning (ML) literacy into a STEM workforce development program for high school students centered on scientific inquiry. SRMPmachine consisted of a 150-hour summer ML Science Institute followed by mentored scientific research internships, positioning ML as a tool for inquiry within the natural sciences. Using a mixed-methods time-series design, we analyzed survey data from 42 high school participants (hereafter, youth or participants) across three timepoints (pre-Institute, post-Institute, post-Internship) alongside retrospective interviews and research artifact analysis from a representative subset of 17 youth. Quantitative results showed significant gains in ML knowledge and skills following the Institute, with particularly strong learning growth among youth from underrepresented groups. While survey measures of attitudes and perceptions toward AI failed to detect any impact of the intervention, interviews suggested nuanced shifts toward what we describe as “informed ambivalence,” characterized by increased ethical awareness, and more differentiated perspectives on AI's societal role. Participants also reported emerging self-efficacy and belonging, describing greater confidence in engaging with AI concepts and communities once they acquired the language and practices of ML. Together, these results suggest that integrating ML learning into authentic science mentorship is associated with gains in AI understanding and youths’ emerging sense of themselves as future AI users and contributors.