2026 ASEE Annual Conference & Exposition

In this paper, we present findings from the second year of a design-based research project preparing autistic community college students for AI and software careers. Building on evaluation cycles that documented persistent challenges with interview anxiety, instructor variability, and the tension between technical skill development and workplace readiness, we report how the program evolved from a "teach CS + teamwork" model to an integrated framework centered on adult-focused social-emotional learning (SEL) and ethical AI.

Qualitative analysis of instructor and mentor interviews revealed that generic UDL-based "communication skills" instruction was insufficient. To make the best use of their new communication skills, students required additional coaching in measurable SEL microskills: self-advocacy under stress, emotional regulation during collaboration, and timing of disclosure decisions. Moreover, instructors struggled to convey communication skills training consistently, exposing a critical gap between pedagogical goals and classroom implementation. This gap motivated the development of a scenario-based train-the-trainer model with structured modules addressing SEL coaching, talk-through facilitation, and inclusive teamwork scaffolds.

The program confronted another emerging challenge: supporting the ethical use of large language models (LLMs) as learning tools without causing academic integrity violations. Our data suggest that for neurodivergent learners, LLMs function as nonjudgmental rehearsal partners for practicing interview responses and debugging explanations. Yet, instructors wished for clear frameworks for distinguishing productive AI support from problematic dependency. This tension reframes AI literacy not merely as a technical policy issue but as an emotional and cognitive support practice requiring explicit ethical decision-making guidance.

In our presentation of this paper, we will discuss formative and summative results from the second year of implementation. We will share instructional resources, student outcomes, and lessons learned about the sustainable implementation of inclusive AI education. Although we have finished our two-year grant-funded work, we will implement a third year of the program as an unfunded extension during Spring 2026 and will report on efforts to sustain the program. We contribute an SEL-integrated design narrative transferable to other computing programs serving neurodivergent learners.