Book review: Ray Kurzweil really, really doesn’t want to die

The inventor and futurist has written “Singularity is Nearer,” a sequel to his 2005 book, which was written when the moment people merge with AI was merely near. He guessed a lot of things right two decades ago, and if he’s just as prescient once again, we’re in for a hell of a ride.

The first person to live 1,000 years has likely already been born. Yes, you heard it right, and it’s not Kurzweil who originally said it – it was Aubrey de Grey, an English biomedical gerontologist, known for his view that technology may enable humans alive today to not die from age-related causes.

Kurzweil agrees with de Grey, obviously, and expands his theories in a new book, Singularity Is Nearer, which is already on the New York Times bestseller list.

The book is popular, of course. After all, Kurzweil predicted quite a bit correctly back in 2005 with his first book about AI and the future of humanity, so now, when he speaks again, people listen.

Besides, Singularity Is Nearer is packed with fantastical visions about how the lives of ordinary folk will massively improve and how these nanobots will fix aging and we’ll all live forever.

As humans, we’re sort of fine with knowing we’re all going to die. But if someone as respected as Kurzweil tells you that’s not necessarily the case, we’re gonna grab this ray of hope and hold onto it.

Alas, even to me, admittedly a tech dilettante who doesn’t like numbers too much, Singularity Is Nearer is truly over-the-top, extravagant, and even naive.

Machines are great when they help out – but merging with them and augmenting our biological brains to alter the course of life as we know it? It will happen – but surely, not for the billions of us.

To be fair, what Kurzweil does is commendable because predictions of a bright future are rare these days – even a solid chunk of AI ideologues warn of possible calamities.

And his arguments aren’t that empty, even when they’re mostly taken from the realms of pop science and general futurism drivel. But among all hopes, false hope is the worst.

Marching, then sprinting

First, though, a dose of optimism. Kurzweil, an excited teenager in the body of a 76-year-old, lists all the awesome advances in AI development and machine learning in particular.

“Something remarkable has happened. Progress has continued to accelerate in defiance of the doubters,” writes Kurzweil who is confident all this is happening thanks to huge growth of computational power.

And indeed, social media and smartphones have gone from virtually nonexistent to all-day companions that now connect a majority of the world’s population. Algorithmic innovations and the emergence of big data have allowed AI to achieve startling breakthroughs sooner than even experts expected.

Computer systems have mastered games like Jeopardy! and Go. They now drive cars, write essays, pass bar exams, and (more usefully) diagnose cancer sooner.

Powerful and flexible large language models like GPT-4 and Gemini can translate natural-language instructions into computer code – dramatically reducing the barrier between humans and machines.

Neural networks have begun unlocking major medical discoveries by simulating biology digitally. We’re even gaining the ability to finally connect computers to brains directly – Elon Musk’s Neuralink is getting ready for a second implant.

“Humanity’s millennia-long march toward the Singularity has become a sprint. These are the most exciting and momentous years in all of history,” says Kurzweil.

He imagines that the moment when a program passes the famous Turing test (a test of a machine's ability to exhibit intelligent behavior equivalent to, or indistinguishable from, that of a human) will come in 2029. Singularity, the day when people will merge with AI, will allegedly follow 15 years later.

Interestingly, the program will actually need to make itself appear far less intelligent in many areas because otherwise, it would be clear to a human tester that it is an AI. That’s because AI programs will be far better than humans in most fields.

The numbers are solid. Kurzweil started in AI in the 1960s, and it has taken sixty years to reach the current level of computation. However, the amount of computation needed to train a new state-of-the-art model is now increasing by about a factor of four each year.

Want another mind-blowing comparison? Here: “One dollar now buys around 600 trillion times as much computing power as it did when the GPS was developed.”

What’s more, AI-level intelligence is truly superhuman-level intelligence – the abilities of the most skilled humans in various domains. Think about it: you might be a mathematical genius but a terrible chess player, right? The AI of the near future, according to Kurzweil, would have no weaknesses.

Nanobots will save us all

So what’s in store for us, then? Longevity or eternal life, Kurzweil says. Bear with me and do blink a couple of times before you read the next quote explaining human life after the Big Merge, though.

“Ultimately, brain-computer interfaces will be essentially noninvasive – which will likely entail harmless nanoscale electrodes inserted into the brain through the bloodstream.”

Kurzweil's title at Google – which is where he is employed – is a bit tone-deaf: “Principal researcher and AI visionary.” The idea of nanobots fixing our brains and stopping aging sounds like something only fetishist visionaries can envision, indeed.

One would think that in the world of super-smart AI, it’d be enough for 3D printers to produce cheap clothing and housing for everyone, or for techies to find an efficient way to grow crops vertically on an industrial scale.

There’s something disconcerting and, frankly, disbelievable about Kurzweil's almost casual writing about nanorobots that will enter our bodies, form a virtual neocortex, and regularly kill all the bad cells, effectively curing cancer.

And of course, innovative new healthcare techniques would be extremely useful. Even (safe) autonomous cars would be amazing – they don’t drink or smoke weed and drive. They can’t be silly.

However, there’s something disconcerting and, frankly, disbelievable about Kurzweil's almost casual writing about nanorobots that will enter our bodies, form a virtual neocortex, and regularly kill all the bad cells, effectively curing cancer.

On the face of it, the argument that we will live longer isn’t that illogical. In 1924, life expectancy in the US averaged about 58.5 years, so babies born that year were statistically expected to die in 1982.

But medicine saw so many improvements during that interval that many of these individuals lived into the 2000s or 2010s. For babies born in 2024, the advances during the years that get added to their lives will be exponentially faster than those of the previous century.

Still, the normal human limit for living is around 120 years – only one person, Jeanne Calment, survived to age 122. Doctors say that even in seniors whose health is fine, the body apparently just starts breaking down – probably due to a combination of cellular metabolism and cellular reproduction.

The solution? Well, to Kurzweil, it’s of course, curing aging itself, and nanorobots – best imagined as “tiny metal robotic submarines chugging through the bloodstream” – will do that. They will repair organs and augment them.

“Eventually, using nanobots for body maintenance and optimization should prevent major diseases from even arising,” writes the author. The little ones will also destroy bacteria and viruses, halt autoimmune reactions, or drill through clogged arteries.

And even if all that doesn’t prevent a human from dying, “in the early 2040s, nanobots will be able to go into a living person’s brain and make a copy of all the data that forms the memories and personality of the original person: You 2. Such an entity would be able to pass a person-specific Turing test and convince someone who knew the individual that it really is that person.”

Threats are there to be shrugged off

“As AI unlocks unprecedented material abundance across countless areas,” Kurzweil writes, “the struggle for physical survival will fade into history.”

He’s certainly not shy. In an interview with The New York Times, Kurzweil admits he is planning to live long enough to experience singularity, merge with AI, and essentially become immortal.

Now, let’s hold our horses, shall we? First, Kurzweil clearly is a techno-optimist, and those can seem myopic and entitled in the face of the world’s many problems. Can we really say AI will solve every problem when we’re not even doing anything about the true causes of those problems?

Kurzweil is just so obsessed with the idea of not dying and selling it to his audience in Silicon Valley that he doesn’t see wider issues at play – and makes some startlingly weird claims.

According to him, no, people’s wages haven’t stagnated over the last forty years in the US (they have) but we shouldn’t worry because we get more bang for our bucks nowadays and because we enjoy so many free things (like scrolling on TikTok or Facebook). Why do we need money anyway, then? Let’s rethink what GDP means. Please, Ray.

Yes, Kurzweil does write about possible and probably inevitable risks to jobs and the potential of super tech to cause mass casualty events – how would you like to live in a society where anyone can build a biological weapon in their kitchen?

But he shrugs the threats off because, guess what, “our civilization will overcome these perils – not because the threats aren’t real but precisely because the stakes are so high.” Also, we will simply need to build fail-safe bots. How? Kurzweil doesn’t know – it’s just too far in the future.

He also writes a lot about extending the exponential curves in computing and exploring, or rather enjoying their spectacular consequences. But to some, this is a fallacy – humans landed on the Moon in 1969, and one could have expected lunar bases and hundreds of other missions to follow, but exploration stalled or even regressed.

Resources are limited – chip wars! – so exponential growth never lasts. It looks and probably is exponential at first but then tapers off and flattens completely. Already today, the AI boom is slowing down – true innovation is out there, of course, but it’s drowning in snake oil.

At least Kurzweil admits that jobs might be automated before new forms of markets emerge and that violence may explode in the interim. Humans might be plagued by a sense of uselessness or inferiority.

Eventually, however, we will stop complaining and rejoice at “our liberation from the burdens of physicality,” says Kurzweil and, naturally, calls the skeptics and the critics “misguided Luddites.” I wouldn’t be so sure – the debate is not settled yet.