Here’s my review of Novacene, by James Lovelock
To mark his 100th birthday on 26 July.
Written for the Literary Review
Few people produce a new book in their 100th year; fewer still at that age produce a book containing original ideas. But if anyone was going to do it, it surely had to be Jim Lovelock. Lovelock has been having good ideas for at least 75 of the past 100 years, and is best known for one that occurred to him half his lifetime ago — the concept of the Earth as a living organism, Gaia. His new book looks forward to the future of that organism, a future in which humankind is unlikely to play a major role, having fulfilled its “purpose” by ushering in an era of artificial intelligence, the Novacene.
I should at this point declare an interest. I have known Jim for more than half my life, and nearly half of his, and have written a biography (now clearly in need of updating!) covering a large part of that life. I come to praise Caesar, not to bury him. That said, however, if his latest book had been the ramblings of a once great mind in its dotage, as a friend I would have ignored it. But because it is as important and accessible as anything he has written, if shorter than one might have hoped, I can recommend it with a clear conscious.
Underpinning Lovelock’s narrative is his conviction that as the harbinger of intelligent life our planet is probably unique, at least in our Galaxy if not in the Universe as a whole. This may seem to fly in the face of the latest discoveries of myriads of planets orbiting stars in our neighbourhood, but is based on a sound assessment of the chain of unlikely events that led our emergence. Life may be common elsewhere without having become intelligent. In emphasising how inimical to life even our near neighbour Mars is, and the likely fate of any human who ventured on to its surface, Lovelock quips “would-be spacefarer Elon Musk has said he would like to die on Mars, although not on impact. Martian conditions suggest death on impact might be preferable”.
The theme running through Lovelock’s book is the way that life has maintained habitable conditions on Earth, unlike those found on Mars, even though the heat output of the Sun has increased. Our role in this has given the name “Anthropocene” to the recent phase of Earth history. His contention is that we are now seeing the beginning of a new age, the Novacene, which will be dominated by hyperintelligences that have evolved from our machines. It all began, he argues, with Thomas Newcomen’s atmospheric steam engine, which ushered in the Industrial Revolution three centuries ago. The key was that these engines could run on their own without the need for constant attention from a human operator, thanks to the feedback mechanism of a regulator that prevented the engines from either running away and exploding or grinding to a halt. Subsequent developments can be seen as evolution at work — evolution proper, not an analogy, as successful designs were (and are) copied, reproduced, and improved. Lovelock reminds us that one description of evolution is “The organism that leaves the most progeny is selected” and says “The steam engine was certainly prolific and so were its successors”.
We have already passed the “Newcomen moment” of the dawning Novacene. Lovelock pinpoints 2015, the year when a computer programme called AlphaGo beat a professional Go player at his own game. It was succeeded by AlphaZero, which turned itself into a formidable chess and Go player by playing games against itself and learning the best techniques through a process of what Lovelock calls “AI intuition”. In 24 hours, starting only with a knowledge of the rules of the game, the machine became a better chess player than any human. As Lovelock says, “we don’t even know exactly how much better it is at any if these games than a human because there are no humans it can compete against.”
You may take comfort in the thought that the programmes were, of course, written by human beings. But that, it turns out, is not a recommendation. Human-written computer code is “the most appalling stuff”, says Lovelock. “It is absolute junk, mainly because it is simply piled on top of earlier code, a shortcut used by coders”. When cyborgs start from scratch, like AlphaZero learning chess, they will start with a blank slate and produce code far superior to ours.
Lovelock has an uncomfortable example of human inadequacy. Modern airliners have computer autopilot systems which can do everything, including takeoff and landing. For safety, these systems are tripled, so that if one system fails the others can carry on. And there are always pilots on board in case all the systems fail. But there is a rare but serious problem with this. Under extreme flying conditions, a situation can occur in which the computer systems do not know what to do. They are programmed in these circumstances to hand control back to the pilots — exactly when conditions are at their worst, and the pilots have been lulled by long experience into trusting the autopilots. Several fatal crashes have occurred when human pilots have “been presented with a problem beyond the capacity of the world’s best autopilots”. The solution might be to get a system like AlphaZero to learn how to fly an airliner by trial and error, although that could work out expensive in aircraft.
Lovelock sees three key events defining the history of life on Earth — or rather, the history of Gaia. The first occurred 3.4 billion years ago, when photosynthetic bacteria first appeared, converting the energy of sunlight into useful form; the second was in 1712 when Newcomen invented an efficient machine to convert solar energy locked in coal into useful work; the third will be when our heirs, the hyperintelligent machines, convert sunlight directly into information. And why should they stop with one star? “Perhaps the final objective of intelligent life is the transformation of the cosmos into information”.
Stated baldly, this sounds like science fantasy, not even science fiction. But even in such a short space, Lovelock, ably assisted by Bryan Appleyard, bolsters his claims with sound scientific reasoning. And what will become of us? He does not envisage a Terminator style conflict between machines and humans. Rather, that their world will be “as difficult for one of us to comprehend as our world is to a dog . . . we will no more be the masters of our creation than our much-loved pet is in charge of us. Perhaps our best option is to think this way, if we want to persist in a newly formed cyber world.”
Like all good showmen, Lovelock leaves his audience thirsting for more. And I wouldn’t put it past him to provide it. Having attended his 90th birthday party and been confidently invited then to reconvene in ten years’ time, I now look forward not only to the next party but to his next decade.
John Gribbin is the author of The Reason Why: The miracle of life on Earth (Penguin, 2012).