f anybody builds it, everybody dies.” Two nonscientists recently wrote
a book with that title, sent it to a bunch of computer experts, and used their
comments to convince us the scientists agreed that artificial intelligence
will cause everyone to die.
In fact, the scientists were probably wondering whether the authors themselves would die of embarrassment when it doesn't happen. Even Science mag showed an unexpected hint of skepticism (good job, Science). It's a demonstration that the best way to disprove your own argument and create skeptics is to predict an inevitable catastrophe.
Are these the same guys who are saying humans will go extinct within ten years from global warming? That the ecosystem is in imminent danger from microplastics and PFAS? That Trump is destroying democracy? And that every book, including Shakespeare's plays, is so traumatizing that people need trigger warnings to avoid getting PTSD?
I can't read minds, but it's hard to believe anyone could seriously think these things. They're only saying it for attention and for political or financial gain. Even if there are some plausible scenarios in the book, which I'll review next week, the title is clearly a non-sequitur. Update: here.
It's also incomplete: everybody dies, and . . . so? What's the downside?
We heard the same scare stories about ChatGPT. People really believed it was intelligent, as the programmers' PR managers claimed, until it practically turned into a joke. Last week the Chicago Sun-Times reported that librarians are being ‘flooded’ with requests for books that don't exist. It turned out that reporters were using some chatbot to produce a list of exciting new books. The only exciting thing about the books was that they didn't actually exist. Over 90% were hallucinated by the chatbot.
In 1910, scientists discovered that the tail of Halley's comet contained cyanogen (N≡C–C≡N), a highly toxic dimer of cyanide. Knowing that cyanide is toxic, this caused widespread panic as the comet approached. French astronomer Nicolas Camile Flammarion predicted that if the comet's tail passed through our atmosphere, it would snuff out all life on Earth. Widespread panic ensued. Some people tried sealing their houses to block the gas. Alas, it couldn't have happened, but that didn't stop the doom-mongers, the hysteria, or the quacks who tried to profit from it.
The scenario is similar. Sooner or later one of the end-of-the-worlders could be right, but death by AI seems very unlikely.
The question, of course, is: is this story true, invented, or filled with fake information from ChatGPT? It scarcely matters. The sheer lunacy of getting information from a chatbot and expecting it to be true has spawned even greater lunacy as people send questionable text to an AI to determine whether it's real. It's like asking a known counterfeiter if the hundred-dollar bill he gave you is authentic.
Now, you might ask a different counterfeiter because he'd be an expert and he'd be motivated to trick you into thinking he's honest, or at least a better counterfeiter. But asking the same one?
If somebody built an AI, it would start a new golden age for science, exploration, and culture. A real AI would be an absolute blast to talk to. Think of having conversations with Einstein, Alan Turing, and Machiavelli in your garage all at once. It would raise your IQ by ten points just being around them.
Maybe it could be dangerous if somebody gave it, say, control of an H-bomb. But if so, the problem isn't the AI. The problem would be that the human would be a moron to do that. The AI would only be supplying the proof.
I get it. People are afraid that machines will be smarter than them, so they attack it. Anybody who ever got called a ‘nerd’, ‘geek’, or ‘four-eyes’ recognizes the pattern. Maybe someday somebody will invent a real AI. I'll bet the authors five bucks the humans live through it.
sep 21 2025, 6:35 am. Text box added sep 22 2025
