George Hotz Thinks We've Turned Intelligence Into a Religion

In tech circles, “intelligence” has quietly become a sacred word. Build a smarter AI, the thinking goes, and every hard problem eventually solves itself. George Hotz — the hacker who jailbroke the first iPhone and later founded self-driving startup comma.ai — thinks that’s a dangerous story to believe. His argument, roughly: we’ve stopped analyzing intelligence and started worshipping it. Let’s take that provocation seriously.

One honest caveat first. This isn’t a raging debate yet. Over the past month, the conversation has been small, anchored almost entirely to a single Hacker News post titled “AI 2040 and the cult of intelligence.” It pulled 118 points and 147 comments — modest by HN standards. But the temperature of the discussion ran hot.

Where the “cult” language comes from

Hotz’s core claim is that Silicon Valley treats intelligence as a master key. The narrative is familiar: once AI gets smart enough, it’ll fix the climate, cure disease, and untangle the economy on our behalf.

The problem is that this reads less like a hypothesis and more like faith. That’s why the word “cult” fits. Cults don’t tolerate doubt about the doctrine. And “intelligence equals salvation” has the same closed shape — once you accept it, disagreement gets hard to voice. Treat raw smartness as the highest good, and you quietly wall off the more important question: is it actually good?

It helps to unpack “intelligence” for a second. What we usually mean is the ability to solve problems. But intelligence doesn’t tell you which problems to solve, or why they matter. That’s the domain of values, not capability. Hotz is prying at exactly that seam.

“Extreme, but not wrong”

The most-upvoted response in the thread cut to it fast: “An extreme example, but he’s not wrong.” The framing may be aggressive, the sentiment goes, but the underlying point lands.

Another comment was shorter and sharper: “Yep, that’s horrifying.” Horrifying how? It was aimed at a future where intelligence concentrates in a very small number of hands. When powerful AI is monopolized by a single company or a single state, that intelligence stops being a tool and becomes power.

Which sets up the third reaction: the idea that anyone will get to own a powerful AI “box” is wildly optimistic. The cynicism here is pointed. A world where individuals freely hold frontier-grade AI probably isn’t coming — regulation, cost, or corporate gatekeeping will put a threshold somewhere, in one form or another.

The real fight: openness versus control

Underneath all of this sits an old tension. Do you open up AI’s power, or lock it down?

One camp argues that concentrating powerful AI in a few hands is the real danger, so the answer is to open it — put it in everyone’s reach. That’s the open-source logic. The other camp argues the opposite: precisely because it’s so powerful, not just anyone should have it.

The trap is that both sides are genuinely persuasive. Open it up and you multiply the risk of misuse. Lock it down and you multiply the risk of concentrated power. The move Hotz calls a “cult” is the impulse to skip this hard choice entirely by assuming the intelligence itself will sort it out. That’s outsourcing your judgment to the machine.

Why this is surfacing now

Notice the number in the title: 2040. This isn’t a debate about next quarter. It’s a look roughly fifteen years down the road.

AI is getting visibly more capable every year. Extend that curve and, by 2040, you could be looking at systems with abilities that are hard to picture today. If that’s true, the story we choose to believe right now helps shape what that future looks like. Do we approach intelligence as something to revere unconditionally, or as a tool to be handled coldly and deliberately? Hotz’s point is that the time to ask is now, while the answer is still open.

And the messenger matters. Hotz isn’t an AI skeptic yelling from the sidelines. He builds self-driving cars; he works with these systems directly. When someone who knows AI this well warns against worshipping it, the provocation gets harder to wave away.

Intelligence is a powerful tool. But the moment you start treating a tool like a god, you surrender the one thing that mattered most — your own judgment about how to use it. Will a smarter AI really build a better world, or just a more powerful something? The question worth sitting with: are you still treating intelligence as a tool, or have you already started to worship it?