AI for U

Ep. 47: Governance and Accessibility: How to Build Trust Around AI in Higher Ed

Episode Summary

Returning guest Joyce Peralta, Manager of Digital Communications at McGill University, joins Brian for a thoughtful conversation about what higher education institutions are getting right and wrong about AI adoption. From governance and accessibility to localization and structured content, Joyce shares how McGill is navigating the realities of implementing AI inside a large, decentralized institution. She explains why AI readiness is really about trust, shared standards, and human judgment, not just new technology. The conversation also explores the future of AI in higher education and why successful adoption will depend on building systems that support both consistency and inclusion.

Episode Notes

Returning guest Joyce Peralta, Manager of Digital Communications at McGill University, joins Brian for a thoughtful conversation about what higher education institutions are getting right and wrong about AI adoption. From governance and accessibility to localization and structured content, Joyce shares how McGill is navigating the realities of implementing AI inside a large, decentralized institution. She explains why AI readiness is really about trust, shared standards, and human judgment, not just new technology. The conversation also explores the future of AI in higher education and why successful adoption will depend on building systems that support both consistency and inclusion.

Join us as we discuss: 

• [2:38] Governance, trust, and creating shared standards across decentralized teams
• [11:34] The localization barrier in AI-assisted translation tools
• [21:19] Why AI is still stuck in an accessibility bottleneck

To hear this interview and many more like it, subscribe on Apple Podcasts, Spotify, or our website, or search for AI for U with Brian Piper in your favorite podcast player.

Episode prompt:

You are a governance and strategy advisor specializing in responsible AI integration within institutional environments.

I'm preparing to roll out a new AI use case in my organization. Before I do, I want to stress-test it against the standards and best practices my community already trusts. Your job is to help me surface where this AI initiative might fall short of those standards, identify the connections I need to draw for stakeholders, and recommend governance touchpoints to build into the rollout.

Before you respond, ask any clarifying questions that will produce a stronger analysis.

The AI Initiative: [DESCRIBE THE AI USE CASE — WHAT IT DOES, WHO USES IT, WHAT IT PRODUCES]

Context:

• Organization type: [PUBLIC/PRIVATE, SIZE, CENTRALIZED OR DECENTRALIZED]
• Functional area: [MARKETING, ADMISSIONS, ADVANCEMENT, IT, ACADEMIC AFFAIRS, ETC.]
• Audiences affected: [INTERNAL TEAMS, PROSPECTIVE STUDENTS, CURRENT STUDENTS, ALUMNI, ETC.]
• Timeline: [WHEN YOU PLAN TO ROLL IT OUT]
• Key stakeholders: [WHO NEEDS TO BUY IN, WHO HAS OVERSIGHT]

Our Existing Standards and Best Practices:

• Accessibility: [SUMMARIZE YOUR ACCESSIBILITY STANDARDS]
• Privacy & Security: [SUMMARIZE YOUR PRIVACY AND DATA HANDLING REQUIREMENTS]
• Brand, Tone & Voice: [SUMMARIZE YOUR BRAND AND EDITORIAL GUIDELINES]
• Usability & User Experience: [SUMMARIZE YOUR UX AND CONTENT STANDARDS]
• Other governance commitments: [E.G., BILINGUAL/MULTILINGUAL CONTENT, DEI, DATA SOVEREIGNTY, RESEARCH ETHICS]

How I'm Currently Planning to Roll This Out: [OUTLINE YOUR ROLLOUT APPROACH — COMMUNICATION, TRAINING, OVERSIGHT, REVIEW]

Now help me work through the following:

1. Standards Alignment: For each existing standard above, where might this AI use case fall short? Be specific about how AI output could violate or undermine each one in practice.
2. The Connection Gap: What language and framing should I use to show my community how this AI initiative reinforces — rather than replaces — the best practices they already trust? Give me phrasing I can use in actual rollout communications.
3. Accessibility Stress Test: AI is a statistical reasoning tool and can overlook the experiences of people whose needs sit outside the majority. Where in this use case is that risk highest, and how do I design against it?
4. Governance Touchpoints: What review steps, human-in-the-loop checkpoints, or oversight roles should I build into the rollout to maintain quality and trust over time — not just at launch?
5. Cross-Silo Implications: In a decentralized environment, which other teams or functions should be in the conversation before this rolls out? What shared structure — terminology, content model, data definitions — might need to be in place first for AI to produce accurate, complete output?
6. Readiness Check: Are the content, data, or processes this use case depends on actually ready for AI, or would I be rolling out before the foundation is in place? Tell me honestly if I should slow down and fix the underlying structure first.

Be direct. I need this stress-tested, not validated.