Harvey Mudd Junior Wins National Prize for Research on AI and Introductory Computing

Harvey Mudd College student Yuan Garcia ’27 won second place for his AI research at the 57th Association for Computing Machinery Technical Symposium on Computer Science Education (SIGCSE), held in February. The recognition is a notable achievement for Harvey Mudd and highlights student-led research on one of the most consequential questions in computing education: how artificial intelligence is shaping student learning in introductory computer science courses.

In his project, “Adaptable Metrics to Assess and Improve Introductory CS,” Garcia examined how student work changes when large language models are introduced in Harvey Mudd’s introductory computing sequence. Garcia worked with Florence Lin ’27, Jenny Ngo ’27, Aidan Deshong ’28 and Edward Donson ’26. The team analyzed more than 1,000 student projects completed between 2018 and 2024 and compared work produced before and after AI tools entered the classroom to better understand how those tools may affect learning outcomes.

The study found measurable differences in introductory coursework. In the first course in the sequence, student code generally became more concise after AI tools were introduced, with statistically significant differences across most of the metrics examined. Projects drawing on AI used a narrower range of course concepts, suggesting greater focus but less integration across topics. In the second, more advanced course, the team found fewer changes overall, with most metrics showing no significant differences apart from an increase in lines of code. Together, those findings suggest AI may play different roles depending on students’ experience level and the goals of a course.

The work also reflects Harvey Mudd’s broader commitment to thoughtful experimentation in computing education. Garcia credited Zach Dodds, Leonhard-Johnson-Rae Professor of Computer Science, with helping create the conditions for the research.

“Harvey Mudd is a very small liberal arts college, which gives us the ability to try new things in ways that larger institutions often can’t,” Garcia said. “Professor Dodds had this amazing idea to introduce AI to introductory CS, and with that came all this interesting data that basically nobody else in the world had, so we thought, why not analyze it and share our findings with everybody else?”

From Lin’s perspective, the project’s importance was rooted in curriculum development and its effect on student learning.

“I started this research because I was interested in curriculum development at an introductory CS level,” Lin said. “Curriculum development plays such a large role in students’ learning outcomes and being able to build a tool that could help with the iterative curriculum development process is why I have continued to stay involved with this type of research.”

Garcia, Lin and the rest of the team are continuing to refine their approach and are working to make their metrics more accessible through a web app, extending their research toward a tool that could support future course design at Harvey Mudd and beyond.