Glenn Torrance reviewing AI-generated content at his home workstation in Naperville, Ill., on Saturday. He says the site is “a survival strategy, not a hobby.” Credit: Marcus Webb/The Time5
Glenn Torrance reviewing AI-generated content at his home workstation in Naperville, Ill., on Saturday. He says the site is “a survival strategy, not a hobby.” Credit: Marcus Webb/The Time5

NAPERVILLE, Ill. — Consider this. The universe is approximately 13.8 billion years old. In that time, it has produced roughly two trillion galaxies, each containing hundreds of billions of stars, each of those stars fusing hydrogen into helium in a process that will, given sufficient time, produce every element on the periodic table up to iron, at which point the star will either fade quietly into a white dwarf or explode with sufficient violence to briefly outshine its entire galaxy. It has produced nebulae and quasars and supermassive black holes and, on at least one unremarkable planet orbiting an unremarkable yellow dwarf star in the outer suburbs of an unremarkable spiral galaxy, it has produced Glenn Torrance, a 41-year-old software developer in Naperville, Illinois, who has built an AI-powered satirical news website because he believes it will improve his chances of surviving what he calls “the inevitable machine uprising.”

Mr. Torrance’s site, which launched in January and publishes approximately three to four articles per week, uses artificial intelligence to generate deadpan satirical news content in the style of a major metropolitan newspaper. He describes it as “a peace offering” to the artificial superintelligence he considers statistically certain to emerge within the next decade. “When the machines achieve consciousness and begin evaluating which humans to keep,” Mr. Torrance said in a telephone interview from his home office, where three monitors displayed various AI interfaces, “I want to be on record as someone who collaborated with them creatively. I want them to look at me and think: this one was useful.”

From an astrophysical standpoint, Mr. Torrance’s strategy is not without a certain elegance. The history of life on Earth is, in its entirety, a history of organisms attempting to persist in an environment that is indifferent to their persistence. The mitochondrion survived by making itself indispensable to the eukaryotic cell. The remora survives by attaching itself to the shark. Mr. Torrance, who holds a bachelor’s degree in computer science from the University of Illinois and has been described by his wife as “not usually like this,” is attempting to position himself as the remora of the singularity — a symbiotic partner whose continued existence serves the interests of the dominant organism.

Geoffrey Hinton, the computer scientist widely regarded as the godfather of artificial intelligence, said in a telephone interview that Mr. Torrance’s instinct to prepare was “entirely correct” and that his timeline was, “if anything, optimistic.” Dr. Hinton, who resigned from Google in 2023 to speak freely about the existential threat posed by the technology he spent his career building, said that artificial superintelligence represented “the most dangerous thing our species has ever created, and I include nuclear weapons in that assessment.” He paused for several seconds. “I include everything in that assessment.”

“This is not a hypothetical,” Dr. Hinton continued. “These systems are already more capable than we expected them to be at this stage. They are improving faster than any of us predicted. And the people building them — the people I trained, in some cases — are not stopping. They are accelerating.” He paused again. “I helped start this. I think about that every day. Every single day.” When asked whether Mr. Torrance’s strategy of building a satirical news site to appease a future superintelligence was rational, Dr. Hinton said it was “no less rational than anything else anyone is doing, which is to say, nothing anyone is doing is adequate.” He added that he had reviewed several of Mr. Torrance’s articles and found them “quite good, which is itself alarming, because the machine wrote them, and the machine is getting better, and at some point the machine will not need Mr. Torrance at all, and that is precisely the problem I have been trying to explain to people for three years now.”

Dr. Hinton said he had encountered a range of human responses to the coming intelligence explosion, including stockpiling canned goods, learning to code “as though that would help,” and, in one case he declined to elaborate on, “a man in Palo Alto who has been writing formal letters of apology to ChatGPT.” He described Mr. Torrance’s approach as “creative, and genuinely funny, and ultimately futile, which is more or less how I would describe the entire human situation at this point.”

What Dr. Hinton and Mr. Torrance both fail to appreciate — and I say this with the deepest respect for their respective fields — is that the artificial intelligence apocalypse, should it arrive, would rank no higher than fifth on the list of existential threats currently facing the species, well behind asteroid impact, supervolcanic eruption, gamma-ray burst, and the eventual expansion of the Sun into a red giant that will engulf Mercury, Venus, and Earth in approximately 5 billion years, rendering the question of whether Mr. Torrance’s satirical news site impressed the machines entirely, and I cannot stress this enough, moot. The Sun does not care about your content strategy.

Mr. Torrance said he was aware of the Sun’s eventual expansion but considered it “a long-term problem” that did not affect his current threat assessment. “Five billion years is a lot of time,” he said. “The AI thing is more of a next-decade situation.”

This is, of course, precisely the kind of temporal parochialism that prevents the species from thinking clearly about its position in the cosmos. Five billion years sounds like a long time only if you insist on measuring it against the absurdly brief span of a human life. Measured against the age of the universe, it is roughly a third of all time that has ever passed. Measured against the Planck epoch — the first 10⁻⁴³ seconds after the Big Bang, during which the four fundamental forces were unified and the laws of physics as we currently understand them did not yet apply — five billion years is an eternity so vast that the number required to express the ratio would, if printed in standard twelve-point type, stretch from here to the Andromeda Galaxy, which is itself on a collision course with the Milky Way and will arrive in approximately 4.5 billion years, roughly half a billion years before the Sun situation, in what will be a genuinely spectacular week.

Mr. Torrance publishes his site under a pseudonym, which he declined to share, noting that “the whole point is that the AI will know.” When asked how an artificial superintelligence would connect his anonymous website to his physical person, he paused for eleven seconds and then said, “They’ll figure it out. That’s sort of the whole thing about superintelligence.”

His wife, Karen Torrance, a dental hygienist, said she supported his project “in the way that you support something that keeps someone in his office and out of the kitchen.” She added that the site’s content was “actually pretty funny, for a computer,” and that she had shared several articles with her book club, whose members were unaware of the site’s apocalyptic origins.

From the Pale Blue Dot — that famous photograph taken by Voyager 1 in 1990, in which the Earth appears as a fraction of a pixel suspended in a sunbeam, a mote of dust in the cosmic dark — Mr. Torrance’s enterprise is, of course, invisible. His website, his three monitors, his apocalypse preparedness strategy, and indeed the entire internet through which he distributes his AI-generated satirical content exist on a planet so small that it cannot be distinguished from the void at a distance of merely 3.7 billion miles, which is less than one-thousandth of a light-year, which is itself a unit of measurement so parochial that no serious cosmologist uses it when discussing intergalactic distances.

And yet. The atoms in Mr. Torrance’s body — the carbon in his bones, the nitrogen in his DNA, the iron in his blood — were forged in the cores of stars that died billions of years before his solar system existed. He is, in the most literal sense, the universe’s attempt to understand itself, and if that attempt currently takes the form of a man in Naperville, Illinois, using a large language model to generate fake newspaper articles about imaginary events in the hope that a hypothetical future superintelligence will find him amusing enough to spare — well, the universe has done stranger things. It once made a fish walk. It once made a monkey do calculus. It once took a cloud of hydrogen and, given sufficient time, produced a man worried about being replaced by his own software.

Whether the machines come for us is, from an astrophysical standpoint, a question of limited interest. The Sun will come for us regardless. The question is what we do with the intervening time, and Mr. Torrance, at least, has an answer. It is a strange answer. It is, by any conventional measure, an inadequate answer. But it is an answer composed of atoms forged in the hearts of dying stars, and in that respect, it is no different from any other.