What I actually think about AI
If you’ve been reading my blog for a while, you’ve probably gotten the impression that I’m an AI true believer. I write about using it to ship faster. I make the case that GitOps is a prerequisite for AI-accelerated device management. I get visibly excited about what it can do.
That’s not wrong, exactly. But it’s not the whole picture.
The honest answer to whether I’m pro-AI or anti-AI is: yes. And no. And it depends on what you mean. I’ve been sitting with that complexity for a while without actually writing it down, so here’s my actual position.
I look at AI the way I look at most powerful tools. A hammer can build a shelter or crack a skull. The internet gave us Wikipedia and also gave us a platform for genocide in Myanmar. Nuclear energy powers cities and vaporized Hiroshima. The technology itself is neither the problem nor the solution. What matters - what has always mattered - is who gets to use it, for what, and who gets to decide.
What I write about on this blog tends toward the optimistic because I genuinely am an optimist. I see AI reducing friction in difficult, repetitive work. I see it enabling people to ship things they couldn’t have shipped alone. I think about the accessibility implications: tools that help people with dyslexia write, that assist people with motor disabilities, that give non-native speakers fluency they’ve spent their whole lives fighting for. That stuff is real, and it genuinely excites me.
But I also see what AI is being used to do right now. I see synthetic content being weaponized to harm real people. I see political processes flooded with AI-generated noise designed to drown out authentic participation. I see the environmental costs - a UN report published this week found that global data centers consumed 448 terawatt-hours of electricity in 2025, with AI driving a fifth of that demand, and that number is projected to more than double by 2030. The associated water footprint is heading toward 9.3 trillion liters. At a moment when we can least afford to be cavalier about either resource. And I see a small number of companies consolidating extraordinary power and capital, with the people building the most capable systems becoming some of the most politically influential humans on the planet. That part terrifies me.
I’ve written before about why the wealth concentration issue is specifically a capitalism problem, not a technology problem. The infrastructure required to train and run frontier models is so expensive that meaningful competition is functionally impossible without access to massive capital. The barrier to entry is measured in billions of dollars and access to specialized chips that are themselves a chokepoint controlled by a handful of companies. That’s not going to fix itself.
I’ve heard a lot of comparisons between where we are now and the dot-com bubble of the early 2000s. The argument goes: too much hype, too much money poured into unproven technology, and most of it will burn. Some of that is probably true. There will be a correction.
But I think the dot-com comparison misses the scale of what’s coming. The internet changed how we access information and how we communicate. AI changes how we think and how we work. Those are different orders of magnitude of disruption. I think history will look back on the dot-com era as a precursor - significant in itself, but not the main event.
The comparison that actually feels right to me is the industrial revolution. Which is both exciting and deeply unsettling.
The industrial revolution generated extraordinary wealth and technological capability. It also produced child labor, factory deaths, mass urban poverty, and social upheaval that took generations to partially resolve. It reshuffled power in ways that the people holding power before it resisted - sometimes violently. The question was never really whether it was going to happen. It was. No amount of organizing against it was going to stop it. The question was who would benefit and who would be crushed in the transition. Those answers took a very long time to sort out, and in a lot of ways we’re still sorting them.
We’re in the early days of something similar. Whether that’s a good thing or a bad thing depends almost entirely on choices we haven’t finished making yet.
I think AI has the potential to eventually get us somewhere genuinely different. A world where the drudge work is actually automated and the gains from that are actually distributed. Where nobody has to choose between keeping the lights on and doing the thing they were made for. I’ve been a Star Trek nerd long enough to have a pretty specific picture of what that looks like, and I’d love to see something like it in my lifetime.
I don’t think I will. The people who hold political and economic power under the current system have enormous incentives to keep that system intact. You don’t get to be a tech oligarch in a post-scarcity world. The road from here to there - if we ever get there at all - is going to be contested every step of the way by people who will frame the dismantling of their power as the dismantling of civilization itself. They’ve done it before. They’ll do it again.
What I do know is that refusing to engage is not a strategy. AI is already embedded in the tools you’re using, the decisions being made about your healthcare and your credit and your job applications. Opting out of understanding it doesn’t protect you from it. It just means you have no hand in shaping what comes next.
I use these tools. I write about them. I try to be honest about what they’re good at and what they’re not, what they can accelerate and what they can’t replace. I try to understand what’s actually happening inside them, not just what the companies building them want you to believe about them. That’s not the same thing as being a true believer. It’s choosing to be in the room where it happens, because the alternative - stepping back while that room fills up with people who only have upside to consider - seems worse.
The vision I keep coming back to isn’t the one you hear most often. Most of the discourse around AI and labor is about productivity - AI will make workers more efficient, AI will free up human time for more value-added tasks. The implicit assumption is that humans are primarily productive units who might get to do slightly less drudge work.
I don’t want AI to make art so that humans have more time to be productive workers. I want AI to be a productive worker so that humans have more time to make art.
That’s the future worth building. I’m not sure we’re on track to build it. But I’m not ready to stop trying.