I'm building a plant growth simulation in Three.js. It's a side project — part curiosity, part craft. I orchestrate the software architecture, cross-reference with how plants actually behave in nature, and slowly shape the simulation into something that feels alive.
Recently I realized I could speed things up. Instead of doing everything myself, I could spin up a team of specialized AI agents — an AI architect, a nature expert, a UX specialist — and let them build the thing together. My role would shift to CEO: set the vision, review the output at the end of the day.
So I tried it. I had been testing growth scenarios by hand — tweaking parameters, observing results, giving feedback to my coding agent. Then I let the agent generate and iterate on scenarios itself. It was faster. The results were better. I had made myself the bottleneck in my own project.
That felt strange. Not because the agent was good — I expected that. But because the logical next step was obvious: identify every role I'm currently filling, replace each one with a specialized agent that does it better, and step back into a pure oversight role.
And then the thought: Am I even needed anymore?
Working with AI agents is a skill we're all still learning. There's a dance between letting them loose and maintaining control. I'm getting better at it. And honestly, the CEO approach — outsourcing to expert agents, reviewing their work — is probably the fastest path. It will win on the productivity spectrum long-term.
But here's what bothered me: sometimes I want to be forced to use my intelligence. When you build something by hand, you understand what you've built. The details are what make it interesting — not to the end user, who doesn't care whether a human or an agent tuned the growth curves, but to me, the builder.
That's craftsmanship. And it's about to become economically irrational.
We all know "Produced in China." It used to mean cheap. Now it often just means produced. The label lost its sting.
I think "Produced by Human" is heading the same way — except in reverse. It won't be a quality seal. It will mean probably worse. Slower, less optimized, more errors. A human touched this, and that's not a selling point.
If human intelligence applied to building things becomes a downgrade, what's left?
Hedonistic consumption? The CEO on a yacht, counting money, watching agents run the company? That's one trajectory. And maybe part of me fears that's where this goes — because I'm a workaholic who loves building things. Take away the building, and I don't know what I'd do with myself.
But I think there's something we're missing.
We'll keep choosing humans where we want human connection. That part I believe. No one wants an AI therapist when they're falling apart at 2am — they want a person who has also fallen apart.
Beyond that though, I think the real answer is that we are spectacularly bad at understanding what it means to be human. We've spent all of history building external things — tools, systems, civilizations. We're world-class at that. But the internal frontier? We know almost nothing.
Jung pioneered the map of the unconscious over a hundred years ago, and we're still at the beginning. The shadow. Individuation. The parts of ourselves we don't even know are running the show. We've barely scratched the surface.
Maybe when AI takes over building, the only frontier left is the one we've been avoiding: understanding ourselves. Not as productivity machines, not as intelligence engines, but as whatever it is that has a soul and a subconscious and no idea what either of those actually are.
That's not hedonism. That might be the hardest work there is.
I don't have a clean ending for this. I'm a developer who just watched an AI do his job better than him in a plant simulation. I'm also a CS grad studying psychology, and I'm starting to think that combination isn't a coincidence.
If intelligence isn't our edge anymore, we need to figure out what is. I suspect the answer isn't in our code. It's in us — in the parts we haven't looked at yet.