LearnAI framework bridges AI skill gap across 5 disciplines at university

A new experience report from arXiv (arXiv:2608.19164) introduces LearnAI, a two-layer framework designed to teach AI-assisted problem solving to mixed-ability learners at a comprehensive teaching university. The authors—Weihao Qu, Ling Zheng, Chris Buzaid, and Daniel Crawford—observed that most institutions respond to generative AI with either broad conceptual workshops or technical CS courses, leaving non-coders and intermediate students with few hands-on options. LearnAI closes that gap with a Wide-Exposure Layer, embedding short AI awareness presentations into existing courses across 18 courses in 5 disciplines, and a Customized Co-Creation Layer offering opt-in one-on-one sessions.

In the customized layer, trained undergraduate tutors guide clients through a 5-Stage Pedagogical Script: Problem Framing, Tool-Task Mapping, Iterative Co-Prompting, Deployment and Verification, and Ethical Reflection. Over two semesters, 35 clients co-created 36 portfolio websites and over 20 deployed web applications. Interviews with five clients and two tutors revealed a consistent mindset shift—participants moved from viewing AI as a passive answer machine to engaging it as a collaborative tool under their own direction. A small paired pre/post AI readiness dataset (N=7) offered preliminary descriptive evidence, while boundary cases included clients who felt overwhelmed and some who deliberately rejected AI. The paper positions LearnAI as a practical, adoptable framework, though it notes the evidence is from a single institution and larger validation is needed.