Most books drop off the radar within a year, and most sequels are disappointing. Ray Kurzweil’s The Singularity Is Nearer is an exception on both counts. From day one a New York Times bestseller, it still tops Amazon’s ranking of sales of books in biotechnology.

In part, the subtitle, When We Merge With AI, explains the book’s success. The wars of religion surrounding artificial intelligence have been raging online since the initially nonprofit OpenAI research organization introduced ChatGPT in November 2022. This generative chatbot software offered the closest approximation to human text and image generation ever available to the public, albeit one prone to propagating misinformation and deceptively concrete made-up “facts.” (To the point: When I asked ChatGPT for my own biography, it listed a PhD dissertation I never wrote at a university I never attended.)

That said, Kurzweil’s personal standing as an innovator in Silicon Valley has a lot to do with the book’s success. According to the jacket-flap biography, his title is “principal researcher and AI visionary” at Google. But his credentials as an inventor of scanning and optical character recognition devices, most notably of technology for converting print text to speech for the blind, are not insignificant. While he never amassed the 11-figure fortune of a Zuckerberg, Gates or Brin, he is a true heir to an earlier and more public-spirited style of Silicon Valley superstardom, a time when Google’s creed was still “Don’t Be Evil.”

Nearer is notable for its difference in tone from Kurzweil’s 2005 bestseller, The Singularity Is Near. Near introduced the singularity concept, the irreversible fusion of the human mind and body with technological systems built on artificial intelligence, biotechnology and nanotechnology. And in many ways, Nearer is an update of the predictions made in Near – for example, the forecast that computers will achieve parity with human intelligence by 2029. But there is a telling difference previewed in the subtitles: When Humans Transcend Biology in the first, When We Merge With AI in the second.

The latter book identifies more closely with a paradigm that has attracted bold scientists for over a hundred years – namely, the idea that humanity can use technology to transcend itself as a new race of superior beings, overcoming the limitations of flesh and blood. It’s a movement known variously as transhumanism and posthumanism. While many faiths imagine a transfigured existence for their followers after death, transhumanists believe in technological self-modification on earth and in space – and often in the ultimate conquest of senescence and death.

Silicon Valley libertarians enamored with transhumanism may be shocked to learn that their spiritual forefather was actually a Marxist, the noted X-ray crystallographer JD Bernal who published the book The World, the Flesh and the Devil fully 95 years ago. Lacking knowledge of the existence and role of DNA, much less nanotechnology, Bernal did not envision modifying the body as much as making it obsolete through what later became fodder for bad science fiction films: the brain in a vat that controlled the world via flashing lights and a spaghetti-tangle of wires

Bernal prophesied the goal of 21st-century transhumanists like Kurzweil. Indeed, Bernal’s vision of a network of superior linked brains is astonishing in its anticipation of the cloud: “Consciousness itself may … vanish in a humanity that has become completely etherealized … becoming masses of atoms in space communicating by radiation.” Expecting popular resistance, he foresaw the newly disembodied scientific elite as “a new species” that would “leave humanity behind.”

Only Connect

Early in Nearer, and bolder in some ways than anything Bernal imagined, Kurzweil outlines the origins of the paradigm that has proved spectacularly successful in achieving recent, discontinuously rapid gains in artificial intelligence: “connectionism.” It is impossible to explain the term properly without a major digression. But Kurzweil does succinctly capture its difference from the rule-based techniques that people use to solve problems as networks of nodes that create intelligence through their structure rather than through their content. Instead of using smart rules, they use dumb nodes that are arranged in a way that extract insight from data itself.

Connectionism was introduced to Kurzweil by a founder of AI, Marvin Minsky, while Kurzweil was still in high school. The idea has made it possible to overcome what seemed to be the insuperable barriers to problem-solving even in a world of Moore’s law, where digital processing power seemed to double like clockwork every 18 months or so.

Connectionism as an “aha!” concept emerged in the 2010s when the hardware – advanced- generation chips housing millions of logic circuits per cubic inch – was finally capable of realizing connectionism’s theoretical potential. It allows massive networks to develop new tools that in turn make innovation easier, a form of positive feedback that makes virtually unlimited progress possible as networks design ever more efficient versions of themselves. Kurzweil thus sticks to his original prediction that artificial intelligence will match and exceed the capacity of the human mind by 2029 – only four years from now!

And that’s just for starters. What Kurzweil dubs the “law of accelerating returns” will also bring spectacular new uses of nanotechnology, another late 20th-century technology that was ahead of its time. In 1986, inspired by an idea of the mathematician John von Neumann, the engineer K. Eric Drexler, published Engines of Creation, a book predicting the potential of a “constructor” machine that could create anything atom-by-atom.

Kurzweil does hedge a bit on his timetable for such a universal machine. He thinks nanotechnologists are on their way to demolishing their physicist skeptics, but he acknowledges that despite progress “it will be at least a decade before the field starts maturing.”

Nanotechnology is essential to realize one of Kurzweil’s favorite and most controversial causes: radical extension of life. He has moved far beyond 1990s visions of human cloning and germline genetic engineering to self-modification independent of genetics and reproduction. Kurzweil predicts a four-step evolution beginning with AI-accelerated therapeutics and continuing with medical nano-robots that can destroy pathogens and prevent future disease and aging. This will lead to “longevity escape velocity” by about 2030, when “we can add more than a year to our remaining life expectancy for each calendar year that passes” – in effect flipping Father Time’s one-way hourglass upside-down.

Welcome To the World Soul

We won’t stop there. Kurzweil predicts that because the mode of neural networks can form a continuum with our biological neocortex, “our thinking will become a hybrid of the biological thinking we are accustomed to and its digital extension,” which will “ultimately predominate.” While this does raise the possibility that the brain will become an adjunct to the cloud rather than vice versa, it also offers the hope – beloved of transhumanists since Bernal’s time – that it will be possible to back up the contents of our biological minds, such as memories of events and emotions, to the cloud. Note that while some transhumanists swear by cryopreservation – freezing bodies after death (or, for the thrifty, just heads) – Kurzweil skips this step to immortality because “diligent people will achieve longevity escape velocity” by 2030.

If Kurzweil is right about the law of accelerating returns, modern economics as we know it, which is rooted in the concept of scarcity, will be obsolete. Today, for example, critics of AI fret about its cost in energy use and the resulting need to turn to nuclear power to light up data centers. Microsoft, for example, has taken it upon itself to resurrect the Three Mile Island nuclear power plant to this end.

But extrapolating trends in the reduction of renewable energy costs (to be turbocharged by artificial intelligence), Kurzweil foresees an era of cheap energy – not to mention cheap materials – thanks to new efficiencies in mining through robotics. “In the 2030s it will be relatively inexpensive to live at a level that is considered luxurious today,” he predicts.

How can this be the case when Kurzweil also forecasts the elimination of demand for all manner of job skills – for example, the humans at the controls of trains, planes and automobiles. In some job categories like bank teller, he explains, employers will continue to value their employees because of their unique role in cultivating relationships with customers. But Kurzweil does acknowledge the potential for social conflict arising from rapid obsolescence of whole skill categories.

His answer is further expansion of the social safety net, culminating in a universal basic income (UBI) that is rid of the opprobrium of freeloading yet is generous enough to support the good life. No job, no problem!

Well, maybe some problems. In the 2024 election, after Nearer’s publication date, a majority of the very workers and their families who would benefit most from the security of a UBI voted for candidates determined to shrink the safety net along with much of the scientific research needed to realize Kurzweil’s nirvana of accelerating returns.

The UBI is widely seen as the darling of libertarians. But it also recalls the Marxist origins of transhumanism in the utopian vision of a future without the alienation of work. As Marx put it in The German Ideology (1845), there is the potential to create a world where one can “hunt in the morning, fish in the afternoon, rear cattle in the evening, criticize after dinner, without ever becoming hunter, fisherman, herdsman or critic.”

What if He’s Right?

In a 1965 essay on a previous generation’s technological prophet, Canadian philosopher Marshall McLuhan, the writer Tom Wolfe posed the now-famous question, “What if he’s right?” It is an even more apt question for Kurzweil, for whom the cloud is destined to be an even more intimate connection of machines to the human neocortex than Wolfe had in mind in his 1964 book, Media: The Extensions of Man.

In Nearer, Kurzweil is predicting the formation of an unprecedented bond with nonhuman intelligence. And at least in this case the phrase “is he right?” has a double meaning. For a start, it can signify ethical correctness, as in whether people have a duty, collectively and individually, to improve themselves beyond the limits that evolution has placed on them.

This version of the old dream of a world-soul making the jump from the transhuman to the posthuman has limited appeal beyond Silicon Valley űbermenschen. But Kurzweil has long been aware of the humanistic critique of the singularity concept and takes pains to cite and respond to it. It’s doubtful, though, that the triumphs of large language AI models like ChatGPT and Anthropic’s Claude will change the minds of detractors.

There is also a technical objection to the singularity idea as the nearly inevitable outcome of advances in technology. Some futurist writers assume that the acceleration of digital processing speed will soon make it possible to solve problems that previous generations considered intractable. Kurzweil reproduces a cartoon of a man at a desk writing a sign that begins “Only a Human Can ...” At his feet, the floor is littered with his previous signs like “Only a Computer Can Compose in the Style of Bach” and “Only Humans Can Pick Stocks.” AI enthusiasts mock humanist naysayers for implying that all the other “only human” capabilities will fail to fall to what Kurzweil calls accelerating returns.

Computer scientists have long recognized the existence of classes of mathematical puzzles like the traveling salesman problem for which no efficient algorithm may exist, which suggests that no acceleration of returns to processing power will yield a solution with current hardware. Indeed, much of internet security relies on what conventional computers still can’t do, especially identifying the prime factors of large numbers. But for the true singularitarian, faith demands that accelerating returns rules – that AI will break through the existing limits on digital engineering.

What Could Go Wrong?

Two more serious problems – plus a minor (but telling) one of my own – that accelerating returns are unlikely to solve have arisen since publication of Nearer. One, which Kurzweil might deny is a problem, is the atrophy of unaided human skills through reliance on AI. A study conducted by researchers at Microsoft and Carnegie Mellon University found that professionals who came to rely on artificial intelligence aids became less capable of analysis unaided.

It’s no revelation that skills atrophy with reliance on technology – for example, when GPS replaces map reading. To technology enthusiasts, that is an old story and a price worth paying for greater productivity. But what happens if the cloud breaks down through some natural or human-made catastrophe? Pessimists are busy countering Kurzweil’s law of accelerating returns (aka LOAR) with what might be called ROAR: risk of atrophied resources.

After several years of impressive growth in the capabilities of large language models like ChatGPT, there is a looming threat to further progress: the scarcity of high-quality texts for training the models, and the subsequent encroachment of derivative and inferior AI-generated text and images called slop. Slop is yet another instance of a problem noted as far back as ancient Greece, when Aristophanes complained about how silver-plated drachmas had replaced the real thing. You may know the problem as Gresham’s Law: bad money drives out good.

In the contemporary context, AI content risks becoming almost unusable after several generations of training, because (so far) artificial intelligence does not appear to be able consistently to distinguish human from AI inputs.

Perhaps the biggest challenge is that AI, for all its conquests, still sometimes breaks down when users assume it will crush the problem. After scanning an encyclopedia yearbook article I had written to be included in a collection of my essays, I found many extraneous hyphens, captions and running heads. Surely, I thought, a site trained on a vast corpus of texts (including, according to one site, pirated versions of my own books) would clean up the text in a flash. It was an abject failure even in obvious cases. Thus, on the road to the singularity, don’t discount the possibility of encountering sinkholes of this kind.

* * *

Kurzweil’s latest book offers a welcome excuse to reflect on the profound changes that, like it or not, technology is bringing us. But in the end, Nearer’s most valuable contribution may be as a counterweight to the gloomy studies of contemporary culture and politics in an era of anti-rational populism. Indeed, Kurzweil’s near certainty that technology is poised to save us from ourselves may be the brightest light in a dreary firmament.