What's Our Problem?

I’ve noticed a pattern in how we talk about AI over the past year.

As I lead Enterprise AI Strategy, evaluating over a hundred AI opportunities across the organization, almost every conversation about governance eventually lands in the exact same place. We talk about the models. We debate security. We worry about hallucinations, autonomous agents, approval gates, and who takes the blame when a system makes a bad call.

Those are all important conversations. They’re also completely missing the point. We act like the problem we need to manage is the AI. I don't think it is. I think the problem is us.

Technologies that actually reshape society rarely do so by just swapping out the old way of doing things. They change us by quietly altering our habits until the old way of thinking no longer feels natural. Electricity didn’t simply replace candles; it changed how long our days were. Smartphones didn’t just replace flip phones; they changed how we remember things, how we navigate, and how we spend moments of boredom. The internet didn’t just give us information; it ruined our patience for waiting for it.

AI is doing the exact same thing. Right now, we still treat it like a tool. We open ChatGPT or Claude because we have a specific problem to solve. We ask a question, we get an answer, we close the tab. The AI feels separate from us.

That separation is disappearing fast. AI is becoming persistent. It remembers context. It follows us across devices. It watches our work, our calendars, our communications, and eventually our environments. Soon, interacting with AI won’t feel like opening an application. It will feel like living with a second layer of cognition that is always on.

When intelligence becomes ambient, the rules change. Every technology changes what humans practice. Calculators changed arithmetic. GPS changed navigation. Search engines changed recall. None of those changes were catastrophic, but none were neutral either. We gained new capabilities, but we also let others fade because they just didn't seem worth the effort anymore. Cognitive scientists have a term for this: "cognitive offloading"—basically, handing off mental effort to an external tool.

AI puts this tradeoff on steroids because it reaches into work that we used to think only humans could do. Writing. Research. Planning. Analysis. Synthesis. Decision support. These aren’t just mechanical chores. They’re the activities many of us have used to sharpen our judgment over an entire career. When we lean on these tools without actively engaging with the work, we start to blindly trust the output. Our own critical thinking skills begin to get lazy.

Look, I spend my days building agentic systems because I believe the upside is enormous. We absolutely should delegate work to machines. But we have to be intentional about which work we delegate.

There’s a massive difference between offloading effort and offloading growth.

Some tasks are totally worth automating because they create zero value for the human doing them. Others look inefficient, but they are actually how expertise is formed. We don’t become good leaders simply by making decisions. We become good leaders by wrestling with uncertainty often enough that our judgment actually improves. If AI quietly removes every productive struggle from our work, we gain a ton of efficiency, but we lose something much harder to get back.

That’s why I think the next chapter of AI governance won’t be about the systems at all.

It will be about us. Not because AI has become dangerous, but because it has become completely ordinary. The organizations that win won’t just ask where an agent should have autonomy. They’ll ask which human capabilities are so fundamental that they should never be allowed to get lazy, no matter how smart the technology gets. They will design workflows that force human verification, critical engagement, and a little bit of productive friction.

Governance isn’t just about controlling intelligent machines.

It’s about deciding what kind of intelligence we want to keep for ourselves.