July 11, 2026
Bot wars, but make it weird
AI 2040 and the Cult of Intelligence
Geohot says super-smart AI won’t magically rule us — and the comments went feral
TLDR: Hotz argues that super-smart AI still has to obey boring real-world limits, so sudden machine takeover fantasies are overblown. Commenters turned that into a messy brawl over freedom, censorship, violent edge cases, and whether ordinary people will ever be allowed to control powerful AI at all.
George Hotz’s new post, AI 2040 and the Cult of Intelligence, basically throws cold water on one of tech’s favorite scary stories: the idea that artificial intelligence will suddenly become godlike and take over everything overnight. His argument is almost hilariously grounded: real life is full of busted parts, delayed shipments, warped chips, and boring physical limits. In other words, no matter how smart the chatbot gets, it still can’t make the boat from China arrive faster. That blunt realism got people talking — and then the comments sprinted straight into chaos.
The biggest fight exploded around Hotz’s argument for local AI: a personal assistant that obeys you, not a corporation or government rulebook. Supporters loved the freedom angle, with one commenter invoking old-school software rights like a battle cry. But critics immediately pounced on his deliberately shocking example about using AI to commit violence, with one user summing up the backlash as, basically, “the problem with AI is that it won’t help you kill your wife?” Ouch. Others took the debate in a darker political direction, warning that future AI could quietly become a tool for censorship, bias, and even “thoughtcrime” scoring if governments or companies control it.
So the vibe was equal parts libertarian fantasy, moral panic, and meme factory. One commenter shrugged it off as “geohot being geohot,” while another delivered the bleakest punchline of all: don’t worry about owning your own powerful AI box — you probably won’t be allowed to.
Key Points
- •The article argues that "hard takeoff" AI scenarios are unrealistic because intelligence does not remove physical and industrial constraints.
- •George Hotz cites his experience building hardware at comma as evidence that real-world systems depend on supply chains, component quality, and manufacturing reliability.
- •The article says fictional AI takeoff narratives rely on nonexistent breakthroughs and cannot bypass the laws of physics.
- •It uses examples such as datacenter construction, international shipping, and chip fabrication to show that many bottlenecks are process- and time-based rather than human-speed-based.
- •The article contrasts centralized AI governance with a preferred model of locally controlled AI that acts directly on behalf of its owner.