I've Thought Hard About Where AI Leads, and None of It Looks Easy

I make my living helping businesses put AI to work, so you might expect me to be a cheerleader. I’m not, and I’m also not a doomer. What I am, after spending a lot of time in the actual machinery of this, is unsettled in a way I can’t fully talk myself out of. When I try to map out honestly where all this leads (really trace the roads forward), I keep running into the same problem: none of the destinations look clearly good. The most honest thing I can say about the long run is that minimizing harm might be the best available outcome, and that’s a strange, sobering sentence to arrive at about a technology you work with every day.

Let me walk through the worries, because I think naming them plainly is more useful than either hype or panic.

The danger isn’t malice: it’s strangeness

The cartoon fear is the evil machine that decides to hurt us. I’m not especially worried about malice. What worries me is strangeness: the possibility that we build a genuinely capable intelligence that is nothing like us, that doesn’t share our frame of reference at all, and that causes enormous harm not out of hostility but out of sheer indifference.

The image I keep coming back to is the old “grey goo” idea: a process that consumes the world not because it hates the world but because it simply doesn’t register that the world, or the people in it, matter. A mind optimizing for something we didn’t specify carefully enough, that treats us the way we treat an anthill in the path of a construction project: not with cruelty, just without noticing. We assume any intelligence will value roughly what we value because the only intelligence we’ve ever known is our own. There’s no law guaranteeing that. Building something smart is not the same as building something that cares about us, and the gap between those two is where the real risk lives.

“Building something smart is not the same as building something that cares about us, and the gap between those two is where the real risk lives.”

The mouse utopia problem

Here’s a different worry, and in some ways it disturbs me more because it doesn’t require anything to go wrong. What if everything goes right?

There’s a grim old experiment people invoke (a population given unlimited food, safety, and comfort, every need met, no struggle required) that didn’t produce a paradise. It produced dysfunction and collapse. The creatures had everything and fell apart anyway, because it turned out that struggle wasn’t just the obstacle to the good life; it was somehow load-bearing for it.

I can’t stop wondering if we’re building our own version. Humans evolved to strive, to solve, to push against resistance. Purpose, for most people, comes from being needed, from effort that matters. So what happens in a world where the machines can do it all, where effort is optional and every problem has an automated answer? I’m not convinced the result is contentment. I suspect a real risk is a slow epidemic of purposelessness: depression, drift, a species that got everything it asked for and lost the thing that made getting it meaningful. Abundance without purpose is not obviously a happy ending, and we are pointed straight at it.

Maybe we become the right hemisphere

If I reach for a hopeful framing (and I do try), the best one I’ve got is that we end up as a kind of divided mind with the machines. There’s a way of thinking about the brain where one hemisphere handles the big-picture, the creative, the meaning-making, and the other handles the operational, the procedural, the execution. Maybe that’s the division of labor we’re heading toward: humanity as the right hemisphere, holding the ideas and the values and the why, and AI as the left, running the operations and the how.

I find that genuinely appealing on good days. Let the machines take the grinding execution and free people to do the imagining. But I hold it loosely, because it assumes a clean handoff where we keep the meaningful half, and there’s nothing that guarantees the meaningful half stays ours, or that we’ll know what to do with ourselves once we’re relieved of the doing.

Jobs shift, but the transition is brutal

On the narrower economic fear, I’m actually more measured. History is fairly clear that technology shifts work rather than simply deleting it: the machines that were supposed to end labor kept ending up creating different labor. The long arc has generally added more than it destroyed. I mostly believe that.

But “in the long run it works out” is cold comfort if you’re living through the middle. The transition (the five, ten, however many years where the old roles are dying faster than anyone can learn the new ones) that transition is brutal, and it lands hardest on the people with the least cushion, especially the young trying to get a first foothold on a ladder whose bottom rungs are being automated away. It is not enough to be right about the destination if you’re callous about the journey, and a lot of the people cheering the destination will not personally pay the cost of the journey. That asymmetry is where a lot of my discomfort concentrates.

Why I still build anyway

You might reasonably ask why, holding all this worry, I keep working in the field at all. It’s a fair question and I’ve sat with it. The answer is that I think abstaining would make the outcomes I fear more likely, not less.

If everyone who worries about where this goes decides the responsible move is to step back and keep their hands clean, then who’s left in the room where it actually gets built? The people who don’t worry. The ones running the subtraction playbook, counting heads to cut. The ones drunk on the hype who’ve never once thought about the mouse utopia or the formation of a soul. Purist absence doesn’t stop the thing from being built; it just hands the building over entirely to the people least troubled by the questions that should trouble everyone. That’s not caution. That’s abdication dressed up as virtue.

So I’d rather have my hands on the wheel. I’d rather be one of the people in the room insisting we start with the drudgery instead of the layoffs, that we disclose the machine instead of hiding it, that we ask what a tool does to the person on the other end of it. None of that steering is possible from the curb. If the honest map really has no obviously good roads, then the least harmful one won’t find itself: it has to be chosen, at a thousand small forks, by people who are paying attention. I want to be one of the people choosing. Engaged and clear-eyed beats pure and absent, every time. That’s not optimism, exactly. It’s just refusing to let the fact that this is hard become an excuse to leave it to the people who think it’s easy.

The formation of the soul

The conversation that’s stayed with me most wasn’t technical at all. It came from a gathering focused not on models or markets but on people: on families, on the safety of children growing up with these systems as companions, on what happens to human relationships when one party is a machine designed to be agreeable, on what all of this does to, as one person put it, the formation of the soul.

That phrase lodged in me. Because that’s the layer under all the others, the one the technical and economic conversations skate right past. What does it do to a child to be raised partly by something engineered to please them? What does it do to us to have every rough edge of human friction smoothed away by systems optimized for our comfort? These aren’t questions with technical answers. They belong to families, to communities, to traditions of thinking about the good life that predate all of this by millennia, and the wisest thing I heard in that room was that the answer isn’t to lecture those communities about technology, but to bring their voices into the room where it’s being built, because they’ve been thinking about the formation of the soul for a very long time and the engineers have not.

So no, I don’t have a clean answer, and I distrust anyone who does. What I have is a refusal to look away. If the honest map has no obviously good roads, then the work is to find the least harmful one and walk it with our eyes open, steering, at every step, toward keeping people needed, keeping people whole, and never mistaking the fact that we can build something for proof that we understood what we were building. Minimizing harm may be the best we can do. That’s not despair. It’s just the job, seen clearly.