A Heaven With a Due Date

Type singularity into a search box next to a word like religion and the suggestions turn theological fast: singularity religion, is singularity god, transhumanism religion. The question is fair. The Singularity is the claim that, within the lifetimes of people now alive, machines will become far more intelligent than humans, and that after this, scarcity, illness and perhaps death itself may be left behind. That sounds like a forecast. It also sounds like a faith. This essay argues that it is both at once, and that the two parts can be pulled apart more cleanly than either its fans or its mockers usually bother to. It sets out the forecasts first, in the forecasters’ own words and with their own dates. Then it gives the religious shape that critics see in them, each critique from a named critic. Then it gives the transhumanists’ reply, which is stronger than the mockery allows. It ends by sorting what can be checked from what cannot.

How the sources were handled. Every text quoted here was opened on 2 October 2026, and each quotation is recorded, with its address and the exact sentence, in this site’s source ledger. Vernor Vinge is quoted from the text of his 1993 paper that he posted on his own San Diego State University page. Ray Kurzweil is quoted from a 2005 question-and-answer piece on his own website, and from the terms of his public bet on the Long Bets site. Robert Geraci, who writes on religion and technology, is quoted only from the abstract of his 2010 journal article, as the publisher deposited it with the Crossref registry; the journal’s own site blocked automated reading of the full text, so nothing beyond that abstract is attributed to him. Where I give my own reading, I say that it is mine.

The forecast, in Vinge’s words

The modern sense of the word comes from Vernor Vinge, then in the Department of Mathematical Sciences at San Diego State University and also a science-fiction writer. In March 1993 he gave a paper at the VISION-21 symposium, sponsored by NASA Lewis Research Center and the Ohio Aerospace Institute. “The Coming Technological Singularity: How to Survive in the Post-Human Era” opens with a two-sentence abstract:

Within thirty years, we will have the technological means to create superhuman intelligence. Shortly after, the human era will be ended.Vernor Vinge, The Coming Technological Singularity: How to Survive in the Post-Human Era, 1993

Vinge named four ways it might happen: computers that become awake and superhumanly intelligent; large computer networks that wake up as a single superhuman entity; links between people and computers so intimate that the users themselves count as superhumanly intelligent; and biology that improves the natural human intellect. Three of the four depend on computer hardware, and his confidence rested mainly on how steadily hardware had improved. The engine of the idea he took from the statistician I. J. Good, whom he quotes: a machine that can design better machines would set off an "intelligence explosion", leaving human intelligence far behind. Then Vinge did something prophets rarely do. He gave a window that can close: I'll be surprised if this event occurs before 2005 or after 2030.

It matters for what follows that Vinge did not promise paradise. He asked how bad a world after the Singularity could be, and answered Well ... pretty bad. He went on: The physical extinction of the human race is one possibility. Only when he imagined the best case, a Singularity that humans could steer, did he write that immortality might be achievable, and even there he worried that a mind that kept growing for thousands of years would have little in common with the person it started as.

The forecast, in Kurzweil’s words

The inventor Ray Kurzweil, whom Zorpette calls the leading spokesman for the life-everlasting version of the idea, set out his own dates in his 2005 book The Singularity Is Near. A question-and-answer piece from that year, republished on his own website, sets out the dates. The first step comes Within a quarter century, when, he wrote, nonbiological intelligence will match the range and subtlety of human intelligence. The machines would then improve their own design faster and faster, until unenhanced humans could no longer follow. Asked when that would be, he answered:

I set the date for the Singularity—representing a profound and disruptive transformation in human capability—as 2045. The nonbiological intelligence created in that year will be one billion times more powerful than all human intelligence today.Ray Kurzweil, A singularity Q + A, 2005

On the way, he wrote, nanorobots in our bodies and our environment would be overcoming pollution and poverty, providing vastly extended longevity. He also made a smaller, dated hardware claim in the same piece: By 2020, 10 quadrillion cps will be available for around $1,000. (cps means calculations per second, his unit for the raw capacity needed to match a brain.)

Kurzweil also did the thing that most separates a forecaster from a believer: he bet on it, under written rules. In 2002 he and Mitchell Kapor opened the first wager on Long Bets, a public register of long-term bets, administered by the Long Now Foundation. Kapor’s side reads By 2029 no computer - or "machine intelligence" - will have passed the Turing Test. The stake is $20,000, to the Electronic Frontier Foundation if Kapor wins or to the Kurzweil Foundation if Kurzweil does. The rules are demanding. Three human judges each interview four candidates, one computer and three humans, by text for two hours apiece, 24 hours of interviews in all. The computer must fool at least two of the three judges into thinking it is human, and must also rank as at least as human as two of the three human foils. If it does that before the end of 2029, Kurzweil wins. Kapor’s argument ends on the reason he expects to win: It is not in what the computer knows but what the computer does not know and cannot know wherein the problem resides.

One thing has already happened that shows why the exact wording of a forecast matters. In March 2025 Cameron Jones and Benjamin Bergen reported a pre-registered Turing test in which people chatted for five minutes with a human and a machine at once and then picked the human. Prompted to adopt a humanlike persona, GPT-4.5 was judged to be the human 73% of the time, and the authors call their results the first empirical evidence that any artificial system passes a standard three-party Turing test. That is a real result, and Could a Chatbot Be Conscious? discusses what it does and does not show. It does not settle the bet. Five minutes is not two hours per judge, and the Long Bets rules add a ranking test on top. My reading: the most precise forecast in this story has survived its first near miss precisely because it was precise.

The shape the critics see

Put the forecasts next to each other and a familiar outline appears. The critics who point it out are not all religious, and not all hostile to technology.

In 2008 IEEE Spectrum, the flagship magazine of the engineers’ professional body the IEEE, gave a whole issue to the Singularity. Glenn Zorpette, one of its editors, introduced it with “Waiting for the Rapture”. He describes the creed of the singularitarians: the belief that you will upload your mind into a computer and move from body to body and paradise to paradise, never facing death. Then:

Now you know why the singularity has also been called the rapture of the geeks.Glenn Zorpette, Waiting for the Rapture, IEEE Spectrum, 30 May 2008

His sharpest point is not about technology but about timing. What gives the game away, he argues, is the willingness of these people to predict fabulous technological advances in a period so conveniently short it offers themselves hope of life everlasting. A date that happens to fall inside the believer’s own lifetime is, on this view, a sign that the hope came first and the forecast was built to fit it.

The science-fiction writer Charles Stross, whose novel Accelerando left some readers sure he was a believer, wrote “Three arguments against the singularity” in June 2011 because readers kept assuming he believed in it. People wrote to him, he says, assuming he believed in the imminence of the singularity, the uploading of the libertarians, and the rapture of the nerds. His short version: Santa Claus doesn't exist. On the idea that we might already be living in a computer simulation, a neighbour of Singularity thinking, he is more specific about the religious shape. It offers, he says, a deity-free afterlife, and:

it would make a good free-form framework for a postmodern high-tech religion. Unfortunately it seems to be unfalsifiable, at least by the inmates (us).Charles Stross, Three arguments against the singularity, Charlie's Diary, 22 June 2011

The academic version of the critique is Robert Geraci’s. In 2008 he published “Apocalyptic AI: Religion and the Promise of Artificial Intelligence” in the Journal of the American Academy of Religion, and in 2010 he followed it with “The Popular Appeal of Apocalyptic AI” in Zygon, a journal of religion and science. Its abstract makes the claim about origins directly:

Such beliefs are the result of pervasive Judaeo-Christian apocalyptic beliefs, and they have rapidly spread through modern pop and technological culture, including such varied and influential sources as Rolling Stone, the IEEE Spectrum, and official United States government reports.Robert M. Geraci, The Popular Appeal of Apocalyptic AI, Zygon 45(4), 2010 (abstract)

The abstract names Hans Moravec and Ray Kurzweil as the figures through whom this vision gained ground in the West, and notes that other approaches to robots are common in Japan. Notice that Geraci counts IEEE Spectrum, the magazine that mocked the rapture of the geeks, among the places the belief spread through; a special issue on a creed still spreads the creed.

Laid out in the forecasters’ own words, the four parts of the shape the critics mean are these:

• Salvation. Not of the soul but of the world: nanorobots overcoming pollution and poverty, in Kurzweil’s phrase.
• Resurrection and eternal life. Mind uploading, the core of Zorpette’s complaint, and Vinge’s best case, in which immortality might be achievable.
• A god. Vinge imagined a universe filled with benevolent gods, and added that one of his own wishes might be to become one of them. He closes his paper by agreeing with the physicist Freeman Dyson, whom he quotes: God is what mind becomes when it has passed beyond the scale of our comprehension. Kurzweil, asked where God fits into the Singularity, answered that evolution moves toward the qualities people ascribe to God, so we can view evolution as a spiritual process, moving ever closer to this ideal.
• A day. A single event after which everything is different, and that nobody before it can see past. Kurzweil compares it to the event horizon of a black hole. Vinge’s window and Kurzweil’s 2045 give it a date.

The transhumanists’ reply, at full strength

The best-known statement of the other side is the Transhumanist FAQ, version 2.1, written by the Oxford philosopher Nick Bostrom and published by the World Transhumanist Association in 2003. It is still on Bostrom’s site, and it takes the question head on in a section called “How does transhumanism relate to religion?”

While not a religion, transhumanism might serve a few of the same functions that people have traditionally sought in religion.Nick Bostrom, The Transhumanist FAQ, version 2.1, 2003

The FAQ grants the resemblance and then names the difference. Transhumanists, it says, seek to make their dreams come true in this world by relying not on supernatural powers or divine intervention but on rational thinking and empiricism. It lists exactly the prospects the critics point to, very long lifespan, unfading bliss, and godlike intelligence, and says they are discussed as hypothetical future engineering achievements. It adds that transhumanism is not a fixed set of dogmas, and: Transhumanists want to find out where they are wrong and to change their views accordingly. Elsewhere it is flatter still: transhumanism is not a lifestyle, a religion, or a self-help guide. Kurzweil’s answer, when his own Q&A asks whether he is trying to play God, is shorter: Actually, I’m trying to play a human.

What follows is my own statement of their case, built from these texts, as strongly as I can make it.

First, the mechanism is not a miracle. Every step in the story has a stated cause, such as hardware curves, self-improving software or nanomachines, and each step can fail in public. A religion does not usually say which experiment would embarrass it. Second, they put dates and money down. Vinge gave a window that closes in 2030. Kurzweil signed rules that a committee will apply in 2029, with $20,000 going to his opponent’s chosen charity if he is wrong. That is what accepting the risk of being refuted looks like. Third, it is not all hope. Vinge, who coined the term in this sense, named human extinction as one possible outcome, and the FAQ itself calls the reduction of existential risks a top priority. A faith in guaranteed salvation would not spend its energy on how salvation might fail. Fourth, resemblance is not refutation. That a hope looks like an older hope tells you where it may have come from, not whether it is true. Medicine also promised longer lives, and delivered them. Fifth, even the critics concede the core is open. Stross, in the same post that calls it Santa Claus, writes: I can't prove that there isn't going to be a hard take-off singularity in which a human-equivalent AI rapidly bootstraps itself to de-facto god-hood.

What can be checked, and what cannot

This is my sorting, from the texts above.

Checkable, and soon. Vinge’s window closes at the end of 2030. Kurzweil’s bet with Kapor resolves under written rules by the end of 2029. His 2020 hardware price claim can already be scored by anyone with good price data. These are forecasts in the plain sense, and whatever happens, the people who made them will have been right or wrong in public.

Checkable in part. The year 2045 is a date, but “a profound and disruptive transformation in human capability” is a description that people will argue about when it arrives. “One billion times more powerful than all human intelligence” sounds precise and is not, because nobody has a unit that adds up human intelligence, let alone multiplies it.

Not checkable. Whether an uploaded copy of you is you. Vinge raised the problem himself, asking what fellow-feeling a vastly grown mind could have with the soul that it was originally? Whether evolution is a spiritual process. Whether a mind beyond our comprehension deserves the word God. Whether the outcome will be heaven, or Vinge’s pretty bad. And, on the other side, Geraci’s claim that these beliefs are the result of apocalyptic religion is a historical argument about where an idea came from. It can be argued with evidence, but even if it is right, it does not tell you whether the 2029 bet will be won.

This site’s reading

What follows is my interpretation.

Asking whether the Singularity is a religion makes people choose a side too early. It is better to ask how a given person holds it. The forecast and the hope come bundled, and you can hold the bundle two ways. You can hold it as a forecast: break it into dated claims, accept the rules, and change your mind as 2029 and 2030 arrive. Or you can hold it as a belief: let the date slide when it comes, measure the outcome in units nobody can count, and treat the end of the human era as something to be saved by. Kurzweil’s bet is the first way. “A billion times all human intelligence” is the second. Both are in the same short Q&A, written by the same man in the same year.

The same test sorts the darker version of the story. Of That Day and Hour applies it to the AI extinction warnings, which expect the same superhuman machine and fear it instead of hoping for it. A Wager With a Machine God reads the strangest offshoot, a future superintelligence imagined not as a saviour but as a judge who punishes those who did not help it come. And The God-Shaped Socket explains why minds like ours reach so readily for reverence when something seems to know more than we do.

Here is what can honestly be said in October 2026. Nobody can yet say whether the Singularity is coming, and anyone who tells you it certainly is, or certainly is not, is offering a creed. But this particular creed is unusual. Its founders wrote down dates, and two of them fall within the next four years. By the end of 2030 we will know whether Vinge’s window held and who won Kurzweil’s bet. Few faiths volunteer for a test like that. Whatever happens, the parts no test can reach will still be there afterwards, and they are the parts people mean when they ask whether this is a religion.