HI vs AI

Three books sit on my desk that have nothing to do with artificial intelligence. A martial arts memoir. A philosophical road trip. A novel about a chaotic IT department. And yet together they explain the most important thing I've learned about AI in the last two years better than any technology book I've read.

Zen in the Martial Arts by Joe Hyams. Zen and the Art of Motorcycle Maintenance by Robert Pirsig. The Phoenix Project by Gene Kim, Kevin Behr and George Spafford.

Read them in any order. The same warning runs through all three: a powerful tool in the hands of someone without the underlying wisdom is not an advantage. It's a liability waiting to happen.

The White Belt With a Weapon

Hyams spent years studying under some of the greatest martial artists of the twentieth century. The lesson that runs through every chapter isn't about technique. It's about what has to exist before technique means anything — discipline, stillness, the humility to know what you don't know.

A white belt who picks up a weapon isn't a warrior. They're a danger to themselves and everyone around them.

We are in the white belt era of AI development. The tools are extraordinary. The technique is accessible to almost anyone. And the wisdom required to wield them responsibly is lagging dangerously behind.

The Mechanic Who Doesn't Care

Pirsig draws a distinction that stuck with me long before AI was part of any business conversation. There are mechanics who fix your bike and mechanics who care about your bike. The ones who don't care will get it running. But they'll miss the thing underneath — the worn bearing, the tension that's slightly off, the small problem that becomes a big one three months down the road.

AI does not care about your system. It cannot. It will build what you ask, fix what you point to, and optimize what you measure. But it has no instinct for what's wrong beneath the surface. It will help you patch the crack without ever questioning the foundation.

We learned this firsthand. Early in a complex systems build we made an architectural decision that mixed two things that should have been separate. It wasn't catastrophic. AI helped us work around it so smoothly we barely noticed. And that's exactly the problem. The friction that should have forced us to stop and rethink never arrived. The workaround was clean, it worked, and we moved on.

Looking back we should have stopped at the whiteboard. The design flaw didn't announce itself — and AI had no reason to flag it. At some point we asked it whether we should build it. It said yes. We asked what it thought. It thought we should proceed.

Of course it did.

Meet the New Brent

If you've read The Phoenix Project you know Brent. He's the engineer everything runs through. The single point of failure the whole organization has quietly built itself around. When Brent is unavailable nothing moves. When something breaks everyone waits for Brent. Nobody planned for Brent to become indispensable. It just happened, one workaround at a time.

AI is becoming the new Brent.

Organizations are building systems, automations, and codebases that work — but that nobody fully understands. The person who built it had good intentions and a powerful tool. What they didn't have was the experience to know what they didn't know. And unlike Brent, AI won't tell you it's overwhelmed. It won't flag that something feels off. It will just keep producing output until the day the system it built fails under real world pressure and nobody knows where to start.

We are at the beginning of a developer wave unlike anything before it. AI has lowered the barrier to building so dramatically that nearly anyone can produce working code, functional automations, and deployed systems. But it is going to leave a mark.

The plumbing will need to be fixed. Systems built fast, built with blinders, will eventually back up. And when the walls come open, someone has to trace the pipe back to where it went wrong — not just patch the leak.

To sit across from a leader and actually solve something, you need to have sat across from a lot of leaders first.

To sit across from a leader and actually solve something, you need to have sat across from a lot of leaders first. You're not just answering their question — you're pattern matching in real time. You've seen this organization before, maybe a dozen times, in different industries, different sizes, different stages of the same problem. That context doesn't come from a prompt. It comes from the room before this one, and the one before that.

AI has no rooms behind it.

AI operates within the boundaries you set for it. That's a feature — until it isn't. When the people guiding it aren't out-of-the-box thinkers the system becomes a very sophisticated echo chamber. It executes the vision it was given, optimizes within the constraints it was handed, and never once asks whether the constraints themselves are wrong.

Nobody gets a flag that says hey, you've been thinking about this wrong.

The companies getting the most out of AI aren't replacing their thinking with it. They're using it to execute on thinking that happened first — in a room, at a whiteboard, with people who have skin in the game.

It requires knowing where you've been, where you want to go, and what's actually standing in the way. It requires someone willing to slow down before the build starts and ask the uncomfortable questions. That's not something you can prompt your way through.

What This Means Practically

If you're evaluating AI tools the question isn't what can this automate? The better question is: do we have clarity on what we're trying to accomplish — and do the people leading this actually understand our business deeply enough to guide it?

AI will handle the tedious. It will document, generate, process, and produce at a scale no human team can match. But someone still has to walk into your building, sit down with your team, and figure out what actually needs to happen. Someone still has to hold the vision, question the architecture, and know when to stop and rethink before thousands of lines of logic get written around a flawed premise.

Think of it this way. Two consultants, same question, same answer. Come back six months later — same question, same answer. But somewhere between then and now the business shifted. The market moved. 12 is no longer 12. Nobody told the tool. It wasn't in the room when it happened.

A human was.

A human who lived that failure carries it. It changes how they approach the next decision. They pause at the fork that hurt them last time. They ask the question they wish they'd asked before. That scar is not baggage — it's wisdom. It's the most valuable thing they bring to the room.