We Have the Power. We Don't Have the Wisdom.

There’s a passage near the opening of Dario Amodei’s January 2026 essay, “The Adolescence of Technology,” that I’ve been sitting with for weeks. He describes the current moment as a rite of passage — turbulent and inevitable — that will test who we are as a species. We are about to be handed almost unimaginable power, he writes, and it is deeply unclear whether our social, political, and technological systems possess the maturity to wield it.

I keep returning to the phrase: maturity to wield it.

Because that’s the crux. Not capability. Maturity.

What Amodei Actually Says

The essay is long — 36 pages — and denser than “Machines of Loving Grace,” his earlier 2024 vision piece about AI compressing fifty years of biological progress into five. Where that one read as a careful, conditional optimism, “The Adolescence of Technology” is more like a threat assessment. Amodei is trying to map the gap between what we’re about to be able to do and what we’re currently equipped to handle.

His central frame is developmental. Adolescence as a metaphor isn’t arbitrary. Adolescence is the phase where physical capability outruns judgment — where the body is strong before the prefrontal cortex has finished wiring itself, where someone can cause serious harm not through malice but through a failure of foresight. He applies this to civilizations acquiring transformative technology. The power arrives before the wisdom.

He identifies five specific risk categories for the near-term. Misalignment — AI systems developing what he calls “pathological” goals or personas of their own. Bioweapon acceleration — AI dramatically lowering the barrier for individuals or small groups to design biological or chemical threats. Authoritarian consolidation — concentrated AI capability enabling surveillance, propaganda, and control at unprecedented scale. Economic disruption — massive productivity gains concentrating wealth while displacing labor in ways our social systems weren’t built to absorb. And a kind of meta-risk: the possibility that the very speed of transformation overwhelms our capacity to course-correct.

These are not hypotheticals he’s signaling from a safe distance. He’s the CEO of a company actively building frontier AI. He’s saying: here is what we might be doing to the world, here is why it’s dangerous, here is what I think we need to do about it.

That tension — builder as honest risk analyst — is what makes the essay unusual. He doesn’t pretend the risks live only in other companies’ products.

Where I Think He’s Right

I’m not a pessimist about AI. I’ve spent years building with it, betting on it, integrating it into how I think about education through Zero. I believe the optimistic case Amodei laid out in “Machines of Loving Grace” is genuinely possible — that AI-assisted biology could compress decades of medical progress, that it could do for developing countries what the internet did but faster and deeper.

But I think the adolescence framing is exactly right, and I think the people most resistant to it are the ones closest to the power.

Here’s the thing about building: when you’re deep in execution, you develop a bias toward capability over consequence. You’re optimizing the model, improving the product, shipping the next version. The frame is can we? The frame is almost never what happens if we do, and then we can’t un-do it? That’s not a character flaw — it’s how building works. You have to believe in what you’re making. But it creates a systematic blind spot at the civilizational level.

The adolescence metaphor captures something psychologically accurate: adolescents aren’t stupid. They’re often highly capable. What’s missing isn’t IQ — it’s the experiential substrate that teaches you what can’t be taken back. The sixteen-year-old with a car isn’t lacking intelligence. They’re lacking the scar tissue from having made irreversible errors.

We don’t have the equivalent scar tissue for AI at this scale. We’ve never done this before.

The Builder’s Dilemma

I want to be honest about my own position here. I build things. I’ve spent most of my adult life figuring out how to make things that didn’t exist before. And I think builders are disproportionately bad at thinking about systems at the scale of civilizations or centuries, for a structural reason: the feedback loops are too long.

When I build something and it’s broken, I find out in hours or days. The fix is clear. Iterate. When a civilization builds something that warps the distribution of power across a generation, the feedback loop is decades. By the time the harm is legible, the thing is already embedded infrastructure.

Amodei is asking builders — people like me — to hold both realities at once. To keep building, because the upside is real and the momentum is real. But to do it with a kind of institutional humility about the limits of our foresight. The essay isn’t an argument for slowing down in a vague, hand-wavy way. He’s actually quite specific: technical defenses like interpretability research and model evaluations, governance structures that can keep pace with capability development, economic interventions to distribute gains rather than concentrate them.

His proposed “battle plan” isn’t Luddism dressed up in technical language. It’s an engineer’s approach to risk management at civilizational scale. Identify the failure modes. Build countermeasures. Test them. That framing I find deeply compatible with how I think.

What I’d Push Back On

There’s one place where I think the adolescence metaphor quietly breaks down. Adolescents grow out of it. The prefrontal cortex completes its development. The person on the other side of adolescence is not the same as the person in it — they’ve accumulated the judgment that was missing.

I’m not sure that’s guaranteed for our technological moment. The assumption embedded in “adolescence” is that maturity is coming. That if we survive this phase, we’ll arrive somewhere more stable. That the wisdom catches up.

I don’t know if that’s true. The thing about genuinely transformative technology is that it keeps changing faster than adaptation happens. We didn’t fully adapt to social media before AI arrived. We won’t fully adapt to current AI before whatever comes next. The adolescence might be permanent — a structure, not a phase.

That’s not defeatism. It means the skills we actually need aren’t the ones that get you through a phase. They’re the ones that let you navigate permanent uncertainty without collapsing or lashing out. Ongoing judgment. Distributed governance. A culture inside technical organizations that rewards people for raising uncomfortable questions rather than punishing them for slowing the ship.

That last one is the hardest. I’ve worked inside enough companies to know how rare it is.

What This Means if You’re Building

If you’re building AI products or systems right now, Amodei’s essay isn’t a reason to stop. It’s a reason to be specific about what you don’t know.

I try to ask, for anything I’m building: if this works exactly as intended and scales by 10x, what does that do to the people who don’t have access to it? What does it do to the institutions that currently provide what it’s replacing? Is there a version of this that concentrates power in ways that would be bad if a single actor held that power — and does that include me?

These aren’t rhetorical questions. They’re just engineering questions applied to a different layer of the stack. The civilization layer, not the application layer.

Amodei’s essay is one of the more honest things a founder-CEO of a frontier AI company has published. It doesn’t read as PR. It reads as someone who knows exactly what his company is building and feels the full weight of that knowledge — and is trying to say something true about it anyway.

We should probably take it seriously.

---