Designing an AI Agent That Refuses to Help (and Why It Works)

Wednesday, September 30, 2026 | 4:30PM–5:30PM MT
Session Type: Poster Session
Delivery Format: Poster Session
In a market where AI makes it easy for anyone to look polished on paper, the harder challenge is helping students figure out what is actually distinct about them. This session shares how we built a student-facing AI agent that starts with who the student actually is, requiring them to articulate their values, strengths, and stories before the tool offers any advising or support. Four iterations taught us what it takes to design AI that respects student agency, including early failures where the agent undermined the very learning it was meant to support.

Presenters

  • Jennifer Baytor

    Director, Career Management Coaching & Programming, Richard Ivey School of Business

  • James Kruck

    Learning Experience Designer, Western University