What Teaching AI at a Community College Taught Me About Enterprise AI Adoption

By day, I work with organizations on IT strategy and AI adoption. In the evening, I teach an AI course for adults at Portland Community College.

My students are retirees, small-business owners, job seekers, and mid-career professionals whose work is changing underneath them. And in the first session, nobody asks me about model parameters, context windows, or licensing tiers. They ask four questions:

- "Is this going to take my job?"

- "Can I trust what it tells me?"

- "Who is responsible when it gets something wrong?"

- "How do I know it isn't biased against people like me?"

Here is what twenty-five years of enterprise IT leadership has taught me: those are exactly the questions your employees are asking about your AI rollout. They are just asking them more quietly — in hallways, in exit interviews, or not at all. And an AI program whose people have stopped asking questions out loud is a program that is quietly failing.

AI adoption rarely stalls on the technology. It stalls on unasked questions. Teaching beginners has sharpened four lessons I now bring back to every executive team I work with.

1. Name the fear first.

If I don't address job anxiety directly in the first class, nothing else I teach lands. People cannot learn a tool they believe is there to replace them. The same is true in an organization. An AI strategy that doesn't begin with "here is what changes for you, and here is what doesn't" will meet a resistance that never appears on a dashboard — until the pilot is already dead. Honesty about impact is not a communications nicety. It is the foundation the whole adoption plan stands on.

2. Start with their problem, not the tool.

The sessions that work best in my classroom start from a real task someone walked in with — a résumé, a budget, a hard email to write, a schedule that won't fit. We learn the technology by using it on something that matters to them. Enterprise rollouts usually invert this: procure the licenses first, then go hunting for use cases. Adoption follows relevance, not procurement. If the first experience someone has with your AI platform doesn't make their actual job easier, there may not be a second experience.

3. Governance works better as a conversation than as a policy.

In class, we build the ground rules together: what we will and won't paste into a chatbot, how we verify what it produces, and where a human being makes the final call. People follow norms they helped write. AI acceptable-use policies handed down from IT, by contrast, mostly get worked around. The organizations getting governance right treat it as something built with the people doing the work — which is also, not coincidentally, the only way governance survives contact with real deadlines.

4. Respect the "beginner" questions. They are the audit.

New users ask the questions executives are often reluctant to raise: Where does my data actually go? Why did it give me that answer? What happens, and who answers for it, when it is confidently wrong? Those are precisely the questions a board, a regulator, an auditor, or a journalist will eventually ask about your AI program. If you cannot answer them in plain language, your program is not governed. It is merely deployed.

I expected teaching to be the straightforward part of my week. It has turned out to be the most honest feedback loop in my professional life — a room full of smart adults with no reason to pretend the technology is working for them, and every reason to say so when it isn't.

If you lead technology, my suggestion is simple: find a room like that, and listen before your next rollout, not after it. The questions you hear will be the ones your own organization is already asking.

-------------------------------

I'm currently exploring my next IT leadership role, focused on AI strategy and governance — the people side of adoption is where I believe these programs are won or lost. If your organization is wrestling with that side of AI, I would welcome the conversation.