James Lovelock at 100 … from AlphaGo to Novacene, and how self-learning machines will be the start of new life
July 26, 2019
James Lovelock will always be associated with one big idea: Gaia. The Oxford English Dictionary defines this as “the global ecosystem, understood to function in the manner of a vast self-regulating organism, in the context of which all living things collectively define and maintain the conditions conducive for life on earth”. It cites the independent scientist as the first to use the term (ancient Greek for Earth) in this way, in 1972.
Today, 26 July 2019, Lovelock celebrates his 100th birthday; his long career has sparkled with ideas.
His first solo letter to Nature — on a new formula for the wax pencils used to mark Petri dishes — was published in 1945. But, unusually for a scientist, books are his medium of choice. He has written or co-authored around a dozen; the latest, Novacene: The Coming Age of Hyperintelligence, is published this month.
Novacene takes the story of Google’s AI-based gaming projects, to explain the profound implications for the future of humanity, and how we will shape that future guided by intelligent machines.
Here is an extract from the new book:
In October 2015, AlphaGo, a computer program developed by Google DeepMind, beat a professional go player. At first glance you may have shrugged and thought, “So what?” Ever since 1997, when IBM’s computer Deep Blue beat Garry Kasparov, the greatest chess player of all time, we have known that computers play these sorts of brain games better than humans.
But you’d be wrong to shrug. AlphaGo used two systems — machine-learning and tree-searching — which combined human input with the machine’s ability to teach itself. This was an enormous step forward, but an even bigger one followed. In 2017, DeepMind announced two successors, AlphaGo Zero and AlphaZero, neither of which used human input. The computer simply played against itself. AlphaZero turned itself into a superhuman chess, go and shogi (otherwise known as Japanese chess) player within 24 hours. Remarkably, this program searched a mere 80,000 positions per second when playing chess; the best conventional program, Stockfish, searches 70m. It was, in other words, not using brute force but some artificial intelligence (AI) form of intuition.
AlphaZero achieved two things: autonomy — it taught itself — and superhuman ability. Nobody expected this to happen so quickly. This was a sign that we have already entered the age I call the Novacene. It now seems probable that a new form of intelligent life will emerge from an AI precursor made by one of us, perhaps from something like AlphaZero. I call these new beings cyborgs.
In theory, this life could think one million times faster than us, but, in practice, the gain of one million times is improbable. A practical difference between the thinking and acting speed of AI and the speed of mammals is about 10,000 times. At the other end of the scale, we act and think about 10,000 times faster than plants. The experience of watching your garden grow gives you some idea of how future AI systems will feel when observing human life.
As the same physical restraints will apply to the cyborgs, however, they will be able to move no faster than us. An intriguing disadvantage for cyborgs is that the rapidity of their thoughts might make long-distance travel exceedingly boring and even perhaps unpleasantly ageing. A flight to Australia would be 10,000 times more boring and disruptive for them than it is for us; for them it would take about 3,000 years.
This new life — for that is what it is — will go far beyond AlphaZero’s autonomy. It will be able to improve and replicate itself. Errors in these processes are corrected as soon as they are found. Natural selection, as described by Darwin, will be replaced by much faster intentional selection. So we must recognise that the evolution of cyborgs may soon pass from our hands.
Already the cosy, convenient devices born from AI that perform the drudgery of housekeeping, accountancy and so on are no longer simply the clever designs of inventors. To a significant extent, they design themselves. I say this seriously because no artisan exists who could by hand construct something as intricate and complex as the central processing chip of your mobile phone.
What is revolutionary about this moment is that the understanders of the future will not be humans, but cyborgs that will have designed and built themselves from the AI systems we have already constructed. These will soon become thousands then millions of times more intelligent than us.
The term “cyborg” was coined by Manfred Clynes and Nathan Kline in 1960. It refers to a cybernetic organism: an organism as self-sufficient as one of us, but made of engineered materials. I like this word and definition because it could apply to anything ranging in size from a micro-organism to a pachyderm, from a microchip to an omnibus. It is now commonly taken to mean an entity that is part-flesh, part-machine. I use it here to emphasise that the new intelligent beings will have arisen, like us, from Darwinian evolution. They will not, at first, be separate from us; indeed, they will be our offspring because the systems we made turned out to be their precursors.
In fact, our new children are already around us. Something as simple as a thermostat already displays a degree of autonomy — it turns the heating on and off. At a much higher level, our mobile phones do a whole range of things without any intervention from us and, when we come to AlphaZero, we see something close to absolute autonomy from its human makers. This process will continue at an ever increasing rate.
We will no longer be the only knowers and understanders of the cosmos, but the process of knowing — the transformation of matter into information — will continue. This is, I believe, a fundamental property of the cosmos. Our mastery of information should be a source of pride, but the gift must be used wisely to help continue the evolution of all life on Earth so that it can cope with the ever-increasing hazards that inevitably threaten us and Gaia. We alone, among the billions of species that have benefited from the flood of energy from the sun, are the ones who evolved with the ability to transmute the flood of photons into bits of information gathered in a way that empowers evolution. Our reward is the opportunity to understand something of the universe and ourselves and to become the parents of a new generation of faster, better knowers.
This, I think, is the prime objective of the cosmos — the conversion of all matter and radiation into information. Thanks to the wonders of the Anthropocene, the age of fire, we have taken the first step. We now stand at a critical moment in this process, the moment when the Anthropocene gives way to the Novacene. The fate of the knowing cosmos hangs upon our response. But the long-term threat to both human and cyborg life on Earth is the exponentially increasing output of heat from the sun.
This is simply the logic of any planet illuminated by a main sequence star. The consequences of solar overheating are already upon us and, but for the regulatory capacity of Gaia, the entire living system, our planet would be moving unstoppably to a state like that of hot Venus now. What saves us is the continuous and sufficient pump-down of carbon dioxide from the atmosphere by land and ocean vegetation.
I don’t think there are intelligent aliens on other planets, but let’s pretend for a moment that there are and they are doing exactly what we are doing at the moment — seeking planets in this so-called habitable zone (also known as the Goldilocks zone because it is neither too hot nor too cold). These alien astronomers would reject Mercury and Venus, which are obviously too close to the sun. But they would also reject Earth, which is also too close. Mars, they will conclude, is the only contender.
Earth absorbs and radiates such a prodigious amount of heat that it cannot possibly be classified as lying within the habitable zone. An alien astronomer viewing the solar system would be obliged to wonder about the anomalous surface temperature of our planet compared with that of Venus. To stay in thermal equilibrium, the Earth must radiate more thermal energy, and it does so at the long wavelengths of infrared. This makes the upper atmosphere at the edge of space hot, but, by the same measure, keeps the Earth’s surface cool.
So I think the zone of habitability idea is flawed because it ignores the possibility that a planet bearing life will tend to modify its environment and climate in a way that favours the life upon it, as ours does. A great deal of time may have been wasted during the search for life elsewhere because of the false assumption that the current environment of the Earth is simply a matter of geological happenstance. The truth is that the Earth’s environment has been massively adapted to sustain habitability. It is life that has controlled the heat from the sun. If you wiped out life entirely from the Earth, it would be impossible to inhabit because it would become far too hot.
So we are made by our star, which provides the energy for life, but we are also threatened by it. This star is a perfectly ordinary, somewhat small, middle-aged cosmic entity — a 5bn-year-old main sequence star. Models of the sun explain how it stays hot by fusing its hydrogen into helium in the ultra-incandescent regions of its interior.
But just as burning coal in oxygen produces carbon dioxide, so fusing hydrogen produces helium. Both carbon dioxide and helium are greenhouse gases: the first warms the Earth, the second warms the sun. This makes the inner regions of the sun hotter and so increases the rate of fusion; the extra heat makes the sun expand and from its greater surface area more heat escapes and warms the Earth. It will continue to increase its output of heat until, in 5bn years’ time, it becomes a red giant star and slowly absorbs the Earth and the inner planets of the solar system.
Provided there is no planetary-scale catastrophe, habitable conditions on Earth for organic life will probably last a further several hundred million years. For electronic life forms, such a time span might seem equivalent to infinity, since they could do so much more than we can in a second of our time. For a while at least, the new electronic life might prefer to collaborate with the organic life that has done (and still does) so much to keep the planet habitable.
By remarkable chance, it happens that the upper temperature limits for both organic and electronic life on planet Earth are almost identical and close to 50C. Electronic life can, in theory, stand much higher temperatures, perhaps as high as 200C. But it could never reach such a temperature on our ocean planet. Above 50C the whole planet moves to an environment that is corrosively destructive. In any event, there will be no point in trying to live at temperatures above 50C. The physical conditions of the Earth at higher temperatures than this would be impossible for all life, including cyborgs.
The intriguing outcome of these considerations is that whatever form of life takes over from us will have the responsibility of sustaining thermostasis with a temperature well below 50C. If I am right about the Gaia hypothesis and the Earth is indeed a self-regulating system, then the continued survival of our species will depend on the acceptance of Gaia by the cyborgs. In their own interests, they will be obliged to join us in the project to keep the planet cool. They will also realise that the available mechanism for achieving this is organic life. This is why I believe the idea of a war between humans and machines or simply the extermination of us by them is highly unlikely. Not because of our imposed rules, but because of their own self-interest, they will be eager to maintain our species as collaborators.
The price we would have to pay for this collaboration is the loss of our status as the most intelligent creatures on Earth. We would remain humans living in human societies and, doubtless, the cyborgs would provide us with an unending source of imaginative and enlightened entertainment. Or we could provide entertainment for them, just as flowers and pets delight us. This might be a little too close for comfort to the world in the film The Matrix, in which humans are kept as energy sources by a machine race that keeps them passive by giving them virtual lives in a virtual world identical to the one from which they were evicted. A future as a battery is not an attractive option.
The point about this future with free-thinking cyborgs unencumbered by human rules is that we can neither guess nor mandate what it will be like in the long term. In the short term, I anticipate collaboration in the sustenance of the Earth as a living planet. But, in the longer term, what if the cyborgs ask themselves: why stay on Earth? The needs of cyborgs are quite different from ours. Oxygen is a nuisance, not a vital necessity. There is far too much water for comfort. Maybe they will decide to move to Mars.
When the Novacene is fully grown and is regulating chemical and physical conditions to keep the Earth habitable for cyborgs, Gaia will be wearing a new inorganic coat. Eventually, organic Gaia will probably die. But just as we do not mourn the passing of our ancestor species, neither, I imagine, will the cyborgs be grief-stricken by the passing of humans.
© James Lovelock 2019. Extracted from Novacene: The Coming Age of Hyperintelligence by James Lovelock (Allen Lane £14.99), published on July 4