Here's how the world will end — a risk expert's guide

Times

20-07-2025

This is not a book for the anxious. It tells of the collapse of empires and the potential for the implosion of human society.
In his marshalling of existential risks the author Luke Kemp deploys apocalyptic prose. 'It is deceptively easy to think about killer drones, AI and nuclear weapons as being separate, isolatable threats,' he writes before delivering the kicker: 'They are not.' Instead Kemp talks about a 'tree of doom' in which hierarchy, inequality and war are central.
This weighty tome aims to emulate the impact of Sapiens: A Brief History of Humankind by Yuval Noah Harari and similar doorstoppers. But that kind of book requires an author to possess exceptional gifts. I find myself wondering how anyone can hold forth authoritatively on such a broad range of topics such as palaeolithic lifestyles, inequality in ancient Rome and the spread of microplastics. The answer, I think, is that you are either a polymath like Felipe Fernández-Armesto, whose book Millennium: A History of the Last Thousand Years succeeded in telling a broad swathe of history, or, like Harari, you are such a great explainer of complex phenomena that people don't look too closely at the details.
Kemp aspires to greater things than his predecessors: an even broader view of how humanity developed and how it soon might end. However, he concedes that the first section, about the emergence of hierarchical societies, contains a lot of educated guesswork. 'We don't know when exactly empires and states came to capture most of the human population … at best, states dominate less than 0.5 per cent of the human timeline.'
In the middle section, about the rise and fall of empires such as imperial China, Kemp is on less shaky ground. There's some great detail here, for example: 'Ten to 25 per cent of men above the age of 17 were deployed for combat in the Roman republic. At its peak Rome had an army of about 400,000 to 500,000 men.' But sic transit gloria mundi — military prowess was not enough to preserve this or other civilisations. Kemp says: 'In every case prior to the collapse [of an empire] there is a slow growth in inequality, oligarchy, corruption and factionalism between elites over the surplus of lootable resources.'
All of this sets the scene for the troubling present, our future dystopia and possible extinction, which forms the final section. Kemp works with the University of Cambridge's Centre for the Study of Existential Risk so this is his bread and butter. He allows for nuance — for different interpretations of the risks arising from climate change, biotech, nuclear weapons and AI — but at the heart of it he believes there is a chance of 'a single gargantuan crash' due to our 'hyperconnected, hyperhomogenous and hyperaccelerated systems'.
•
At this point you can almost imagine Kemp's editor urging him: 'For God's sake, give the reader something to live for!' And he tries. 'Most of the challenges we face are entirely solvable,' he writes. His suggested remedies include banning nuclear weapons, addressing climate change, creating more equal societies and limiting AI.
In a section bravely titled Don't Be a Dick! he argues that we can all make a positive difference by eating less meat, turning down certain jobs and shunning particular types of politics. Yet a few chapters earlier Kemp argues that the idea 'we are all to blame … is a manufactured distraction. The reality is that a mere handful of giant corporations, countries and militaries are responsible for the great majority of catastrophic risk.' So are we individually empowered or not? Here's where Kemp's thesis begins to fray badly because not all of us live in societies where change can be brought about democratically.
If you ask me which country is stripping the world's resources at an unprecedented rate, has the biggest state funding of AI, is most rapidly increasing its stock of nuclear weapons and is home to numerous laboratories containing lethal pathogens, the answer is the same: China. But modern China is hardly mentioned in this book.
Instead Kemp's focus is almost exclusively on the US. I don't doubt there are reasons to watch Google or the Pentagon closely, but with some shareholder and democratic accountability, I fancy we have a better chance of finding out what's going on in these places than we do in some Chinese AI project or biolab. This lack of interest in the sources of risk outside the western world extends to countries like India and Brazil.
A book covering this great swathe of time, societal models, scientific progress and risk requires an Olympian detachment, the ultimate long view. But instead what the reader ultimately gets are the comforting 'progressive' orthodoxies of the 21st-century western academic world.