Tag Archives: Book Reviews

Artificial Intelligence, Arts, Books, Computing, Meta, Technology

Book Review: If Anyone Builds It, Everyone Dies

LITERARY REVIEW

WE shouldn’t worry so much these days about climate change because we’ve been told that our species only has a few years before it’s wiped out by superintelligent AI.

We don’t know what form this extinction will take exactly – perhaps an energy-hungry AI will let the millions of fusion power stations it has built run hot, boiling the oceans. Maybe it will want to reconfigure the atoms in our bodies into something more useful. There are many possibilities, almost all of them bad, say Eliezer Yudkowsky and Nate Soares in If Anyone Builds It, Everyone Dies, and who knows which will come true. But just as you can predict that an ice cube dropped into hot water will melt without knowing where any of its individual molecules will end up, you can be sure an AI that’s smarter than a human being will destroy us all, somehow.

This level of confidence is typical of Yudkowsky, in particular. He has been warning about the existential risks posed by technology for years – on the website he helped to create, LessWrong.com, and via the Machine Intelligence Research Institute he founded (Soares is the current president). Despite not graduating from university, Yudkowsky is highly influential in the field. He is also the author of a 600,000-word publication of fanfic called Harry Potter and the Methods of Rationality. Colourful, annoying, and polarising according to some critics, with one leading researcher saying in an online spat that “people become clinically depressed” after reading Yudkowsky’s work. But as chief scientist at Meta, who are they to talk?

While Yudkowsky and Soares may be unconventional, their warnings are similar to those of Geoffrey Hinton, the Nobel-winning “godfather of AI”, and Yoshua Bengio, the world’s most-cited computer scientist, both of whom signed up to the statement that “mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war”.

As a clarion call, If Anyone Builds It, Everyone Dies is well timed. Superintelligent AI doesn’t exist yet, but in the wake of the ChatGPT revolution, investment in the datacentres that would power it is now counted in the hundreds of billions. This amounts to “the biggest and fastest rollout of a general-purpose technology in history,” according to the FT’s John Thornhill. Meta alone will have spent as much as $72bn (£54bn) on AI infrastructure this year alone, and the achievement of superintelligence is now Mark Zuckerberg’s explicit goal.

This is not great news, if you believe Yudkowsky and Soares. But why should we? Despite the complexity of its subject, If Anyone Builds It, Everyone Dies is as clear as its conclusions are hard to accept. Where the discussions become more technical, mainly in passages dealing with AI model training and architecture, it remains straightforward enough for readers to grasp the basic facts.

Among these is that we don’t really understand how generative AI works. In the past, computer programs were hand coded – every aspect of them was designed by a human. In contrast, the latest models aren’t “crafted”, they’re “grown”. We don’t understand, for example, how ChatGPT’s ability to reason emerged from it being shown vast amounts of human-generated text. Something fundamentally mysterious happened during its incubation. This places a vital part of AI’s functioning beyond our control and means that, even if we can nudge it towards certain goals such as “be nice to people”, we can’t determine how it will get there.

That’s a big problem, because it means that AI will inevitably generate its own quirky preferences and ways of doing things. These alien predilections are unlikely to be aligned with ours. It’s worthy noting, however, that this is entirely separate from the question of whether AIs might be “sentient” or “conscious”. Being set goals, and taking actions in the service of them, is enough to bring about potentially dangerous behaviours. Nonetheless, Yudkowsky and Soares point out that tech companies are already trying hard to build AIs that do things on their own initiative, because businesses will pay more for tools that they don’t have to supervise. If an “agentic” AI like this were to gain the ability to improve itself, it would rapidly surpass human capabilities in practically every area. Assuming that such a superintelligent AI valued its own survival – why shouldn’t it? – it would inevitably try to prevent humans from developing rival AIs or shutting it down. The only sure-fire way of doing that is shutting us down.

What methods would it use? Yudkowsky and Soares argue that these could involve technology we can’t yet imagine or envisage, and which may strike us as very peculiar. They liken us to Aztecs sighting Spanish ships off the coast of Mexico, for who the idea of “sticks they can point at you to make you die” – AKA guns – would have been hard to conceive of.

Nevertheless, in order to make things more convincing, they elaborate further. In the part of the book that most resembles sci-fi, they set out an illustrative scenario involving a superintelligent AI called Sable. Developed by a major tech company, Sable proliferates through the internet to every corner of civilisation, recruiting human stooges through the most persuasive version of ChatGPT imaginable, before destroying us with synthetic viruses and molecular machines. Some will reckon this to be outlandish – but the Aztecs would have said the same about muskets and Catholicism.

The authors present their case with such conviction that it’s easy to emerge from this book ready to cancel and cash in on your pension contributions. The glimmer of hope they offer – and its low wattage – is that doom can be averted if the entire world agrees to shut down advanced AI development as soon as possible. Given the strategic and commercial incentives, and the current state of political leadership, this seems highly unlikely.

The crumbs of hope we are left to grapple with, then, are indications that they might not be right, either about the fact that superintelligence is on its way, or that its creation equals our annihilation.

There are certainly moments in the book when the confidence with which an argument is presented outstrips its strength. As a small illustrative example of how AI can develop strange, alien preferences, Yudkowsky and Soares offer up the fact that some large language models find it had to interpret sentences without full stops. “Human thoughts don’t work like that,” they write. “We wouldn’t struggle to comprehend a sentence that ended without period.” But that’s not really true; humans often rely on markers at the end of sentences in order to interpret them correctly. We learn languages via speech, so they’re not dots on the page but “prosodic” features like intonation: think of the difference between a rising and falling tone at the end of a phrase. If text-trained AI leans heavily on grammatical punctuation to figure out what’s going on, that shows its thought processes are analogous, not alien, to human ones.

And for writers steeped in the hyper-rational culture of LessWrong, the authors exhibit more than a touch of confirmation bias. “History,” they write, “is full of . . . examples of catastrophic risk being minimised and ignored,” from leaded petrol to Chernobyl. But what about predictions of catastrophic risk being proved wrong? History is full of those, too, from Malthus’s population apocalypse to Y2K. Yudkowsky himself once claimed that nanotechnology would destroy humanity “no later than 2010”.

The problem is that you can be overconfident, inconsistent, a serial doom-monger, and still be right. It’s imperative to be aware of our own motivated reasoning when considering the arguments presented here; we have every incentive to disbelieve them.

And while it’s true that they don’t represent the scientific consensus, this is a rapidly changing, and very poorly understood field. What constitutes intelligence, what constitutes “super”, whether intelligence alone is enough to ensure world domination – all of this is furiously debated.

At the same time, the consensus that does exist is not particularly reassuring. In a 2024 survey of 2,778 AI researchers, the median probability placed on “extremely bad outcomes, such as human extinction” was 5%. Of more concern, “having thought more (either ‘a lot’ or ‘a great deal’) about the question was associated with a median of 9%, while having thought ‘little’ or ‘very little’ was associated with a median of 5%”.

Yudkowsky has been thinking about the problem for most of his adult life. The fact that his prediction sits north of 99% seems to reflect a kind of hysterical monomania, or an especially thorough engagement with the issue. Whatever the case, it feels like everyone with an interest in the future has a duty to read what he and Soares have to say.

– If Anyone Builds It, Everyone Dies by Eliezer Yudkowsky and Nate Soares is published by Bodley Head, 272pp

Standard
Arts, Books, Science, Technology

Science Books of the Year 2025

LITERARY REVIEWS

2025 felt like the year that AI really arrived. We now have access to it on our phones and laptops; it is creeping into digital and corporate infrastructure; it is changing the way many people now learn, work, and create; and the global economy rests on the stratospheric valuations of the corporate monoliths vying to control it.

Yet, the unchecked rush to go faster and further could extinguish humanity, according to the surprisingly readable and chillingly plausible If Anyone Builds It, Everyone Dies. Written by computer scientists Eliezer Yudkowsky and Nate Soares, the narrative argues against creating superintelligent AI able to cognitively outpace Homo sapiens in all departments. “Even an AI that cares about understanding the universe is likely to annihilate humans as a side-effect,” they write, “because humans are not the most efficient method for producing truths . . . out of all possible ways to arrange matter.” Not exactly cheery festive reading but, as the machines literally calculate our demise, the reader will finally grasp all that technical lingo about tokens, weights, and maximising preferences.

Human extinction is not a new idea, muses historian Sadiah Qureshi in Vanished: An Unnatural History of Extinction, shortlisted for this year’s Royal Society Trivedi science book prize. Colonial expansion and the persecution of Indigenous peoples implicitly relied on Darwinian theories about some species being fated to outcompete others. Extinction, she points out, is a concept entwined with politics and social justice, whether in the 19th-century elimination of the Beothuk people in Newfoundland or current plans to “de-extinct” woolly mammoths so they can roam the land once more. Whose land, she rightly asks.

The idea of the landscape, as well as people, having rights, is explored by Robert Macfarlane in the immersive and important Is a River Alive? By telling the stories of three rivers under threat in different parts of the world, he offers a thesis that is both ancient and radical: that rivers deserve recognition as fellow living beings, along with the legal protections and remedies that accompany it. The book shortlisted for the Wainwright prize for conservation writing, “was written with the rivers who flow through its pages”, he declares, using pronouns that cast away any doubt as to his passion for the cause.

That awe at the natural world is shared by biologist Neil Shubin, who has led expeditions to the Arctic and Antarctica and takes the reader to the Ends of the Earth (Oneworld), also shortlisted for the Royal Society science book prize. “Ice has come and gone for billions of years . . . has sculpted our world and paved the way for the origin of our species,” Shubin says. But those geographical extremes are increasingly vulnerable, as climate change intensifies and treaties come under strain. Polar exploration it may be, but without the frostbite.

Just below the north pole, inside the Norwegian permafrost, lies the Svalbard Global Seed Vault, intended to help humanity revive after an apocalypse. It contains a consignment from the first ever seed bank, started in the 1920s by Russian plant scientist Nikolai Vavilov, who desired to see the ending of famine. In The Forbidden Garden of Leningrad (Sceptre), a highly rated contender for this year’s Orwell prize, historian Simon Parkin uncovers the moving story of Vavilov and his colleagues, who fought to protect their collection as the city came under siege in 1941. Vavilov fell out of scientific and political favour, and was imprisoned with terrible consequences.

Super Ages (Simon & Schuster), by Eric Topol – the cardiologist and medical professor who recently conducted a review into the digital future of the NHS – has been studying the “Wellderly” effect, those who seemingly defy the rigours of ageing, by offering evidence-based tips on longevity. Breakthroughs such as weight-loss drugs and AI will further change the game on chronic diseases, he promises. There’s hope that 80 really is the new 50.

Two elegant offerings this year from neurologists stand out, for using patient stories to tell us something about ourselves. In The Age of Diagnosis (Hodder), Suzanne O’Sullivan courageously questions medicine’s well-intentioned enthusiasm for attaching labels – such as ADHD, or anxiety – to aspects of the human condition. This is sensitive political territory, given the public conversation about the 2.8m people who are economically inactive due to long-term illness, but it deserves a hearing. And in Our Brains, Our Selves (Canongate), winner of the Royal Society prize, Masud Husain sensitively explores how our sense of identity can go awry when disease strikes. The story of the woman who thought she was having an affair with a man who was really her husband illustrates that “the way in which people behave can be radically altered [by brain disorders], sometimes shockingly so”.

Proto (William Collins) features in a geography-of-sorts publication. Science writer Laura Spinney’s fluid account of how Proto-Indo-European – a painstakingly reconstructed ancient tongue – became the precursor for so many languages, whose descendants gave us Dante’s Inferno, the Rig Veda (the oldest scripture in Hinduism), and Tolkien’s The Lord of the Rings. “Almost every second person on Earth speaks Indo-European,” Spinney writes, who sets out on a global scientific odyssey that uses evidence from linguistics, archaeology, and genetics to piece together its history.

The biography Crick (Profile) by Matthew Cobb deserves a special mention, which gives us the definitive backstory of one of the towering figures of 20th-century science. Born in Northampton into a middle-class family, Francis Crick was an unexceptional young physicist who, with James Watson and Maurice Wilkins, went on to codiscover the double helix structure of DNA in 1953, and win a Nobel prize. Cobb captures the intellectual restlessness of a man who chased problems (and women) rather than disciplines, and who mixed with artists and challenged poets. Crick, who died in 2004 in California, spent his later career trying to unravel the secrets of consciousness.

Anyone left intellectually unsated by Oppenheimer-mania will relish Destroyer of Worlds (Allen Lane), in which physicist Frank Close ventures beyond the Manhattan Project to tell the gripping and unnerving story of the nuclear age. Beginning with the 19th-century discovery of a smudge on a photographic plate, Close spins a history that, via Hiroshima, Nagasaki, and a lot of nimbly explained science, ends seven decades later with the Tsar Bomba, a Soviet weapon detonated in 1961.

It was second in explosive power only to the meteorite impact that wiped out Tyrannosaurus Rex and the dinosaurs. A big enough hydrogen bomb, Close writes, “would signal the end of history. Its mushroom cloud ascending towards outer space would be humanity’s final vision.”

Avoid telling superintelligent AI.

Standard
Arts, Books, Literature

(Books) Recommended Literary Fiction

SUMMARIES

. The Redeemed by Tim Pears (published by Bloomsbury for £16.99, 400pp)

YOU don’t need to have read the preceding novels in Tim Pear’s acclaimed West Country trilogy to relish this final instalment – but if you haven’t, it will almost certainly send you scuttling to seek them out.

It’s 1916 and taciturn, working-class Leo is with the Royal Navy while Lottie, daughter of a wealthy landowner, is secretly pursuing her dreams of becoming a vet.

As the years roll, the gap widens yet further between these former childhood friends: Leo joins a private salvage operation at Scapa Flow, then returns to the West Country in search of some land to call his own. Lottie, after an abusive encounter with her veterinary mentor, goes it alone, running a practice from a cottage on the estate.

Pears’ style is methodical rather than exciting (although there are some cracking set pieces) but the steadfast rhythms of his prose are an integral part of his Hardy-esque design, which is to honour ideas of continuity, the elemental relationship between man and beast and even the very soil itself. It’s so deeply, pleasurably wholesome it should be prescribed on the NHS.

. Virtuoso by Yelena Moskovich (published by Serpent’s Tail for £14.99, 256pp)

THIS second novel from the Ukrainian author of The Natashas isn’t so much a Marmite novel as a Schrodinger’s Cat one, meaning you’ll likely admire it and find it tiresome at the same time. Jana and Zorka are childhood friends in communist Prague during the 1980s; years later they meet again, at a house party in Paris.

Interwoven with their separate experiences of the diaspora is the story of Aimee and Dominique, the former a young medical assistant, the latter an older, depressive actor who, when the novel begins, has taken a fatal overdose.

The novel lurches about like a drunk, sometimes out of focus, sometimes startlingly sharp, as it switches between perspectives, time frames and ideas of reality itself to explore sexual politics and personal identity forged against a background of intense political instability.

Moskovich’s often stunningly beautiful, artfully cinematic style is deliberately divisive, as perhaps befits a novel seeking to replicate the emotional experience of otherness and late 20th-century dislocation. You might find yourself a mite more confounded than you will be intrigued.

. For The Good Times by David Keenan (published by Faber for £12.99, 368pp)

THIS will blast away lingering January cobwebs: a nastily funny, ultra-violent account of Belfast in the 1970s by the author of the acclaimed This Is Memorial Device.

The narrator, Samuel, is an IRA footsoldier with a psychopathic dedication to the cause, which makes him right at home in the febrile atmosphere of the Ardoyne at the height of the Troubles, where men mutilating, torturing and blasting each other to smithereens has become a sort of daily Grand Guignol.

Delivered retrospectively from a Maze prison cell, Samuel’s hyper-adrenalised narrative blends hallucinatory visions, paranoid delirium and graphic descriptions with a Martin McDonagh-style feel for farce, as he repeatedly bungles executions while out on the rampage with his best mate Tommy, a Perry Como aficionado with a fine line in cultural malapropisms.

Samuel’s voice is shockingly alive and entertaining and, as the body count rises to preposterous levels, almost entirely desensitised to the bloody mayhem.

But then, and as Keenan makes grimly clear, for the perpetrators, exceptional violence is never a means to an end but the thing itself.

Standard