The Blog

Why the Terminator Is Not Coming Any Time Soon

As profound and exciting success in either the origin of life or artificial life would be for some, it is also scary and terrifying for others. Just recently, figures as notable as Elon Musk and Stephen Hawking openly shared their worries. Could AI pose a threat to humanity?

Spontaneous novelty, artificial intelligence and the origin of life this is why we're not yet facing the robot apocalypse.

How did life start on planet Earth? This is a complex and fascinating question. The fact is, we know more about the molecule-by-molecule workings of biology and the outer reaches of our Universe than the origin of life. The central problem is rather sobering. We will never know for sure how life got started, as there was no one there to observe it. This is also true for the jump from single cell creatures to plants and animals, not to mention the evolution of intelligent creatures who are able to ultimately make a model of their own environment and innovate.

The question of the origin of life on Earth is profound. If we can (re-)create life as we know it, or similar, then it will tell us about how easy it was for life to emerge on planet Earth, and also understand the likelihood of finding life elsewhere in our Universe. People are actively researching many aspects of the origins of life. From a biochemical point-of-view, scientists are looking at the nuts and bolts of biology and trying to see how simpler versions could have started. Biology today contains three basic parts: a biological hard drive containing all the information to build and run a cell, machinery for constructing cells and operating them, and a process of energy storage and conversion. These systems are incredibly intricate and the likelihood of them bursting into existence together - and forming a viable cell - is vanishingly low. Despite many promising avenues of research work, a deep understanding of the process that can transform 'dead' chemistry into 'living' biology remains as elusive as ever.

Other scientists in a variety of fields, from computer science to robotics, are taking a different approach, such as trying to build computer models of living systems. The idea is to see if it is possible to make a computer program simulate life, or itself spring to life. This is a mature field, with many exciting developments in algorithms and robotics. Internet searches, toy drones, and smart(er) cars have all been made possible by these advances. Some have seized the ultimate challenge, which is to build a robot that can reproduce itself automatically, given some raw-material inputs and an instruction code. Success here would yield artificial-life (ALIFE) forms and perhaps ultimately a form of artificial intelligence (AI) but try as they might, the spark of life remains elusive here also.

As profound and exciting success in either the origin of life or artificial life would be for some, it is also scary and terrifying for others. Just recently, figures as notable as Elon Musk and Stephen Hawking openly shared their worries. Could AI pose a threat to humanity? Is the Cyberdyne and the Terminator a real possibility? Could the machines (mechanical or new versions of biology) be a threat to humanity? To me the worries of Musk and Hawking remind me of the comments made by Prince Charles when he voiced his worries over grey goo. The idea of grey goo came from the Michael Crichton science fiction novel 'Prey' which, although a gripping tale, bears little resemblance to any synthetic technology so far produced.

I think we are very unlikely to be facing anything dangerous or sentient in the near future. This is because progress is also slow in computer science and robotics. There is little evidence to suppose that ALIFE or AI are any closer today than a decade ago. Computers are getting faster and 'smarter', but this is simply the result of new programming techniques, algorithms, and speed of information gathering. Could lack of progress in both origin of life, ALIFE and AI be related? I would argue they are deeply connected and understanding the same feature of reality is required to boot up either biology or ALIFE and AI from scratch.

I am going to call this missing feature 'Spontaneous Novelty'.

What is 'Spontaneous Novelty'? I think 'Spontaneous Novelty' is to the origin of life and artificial life what the 'Big Bang' is to cosmology. It is that universal something for nothing. A transition whereby fluctuations on the very small scale that normally cancel out are amplified. One well known example is the case of spontaneous resolution of molecules that have a left or a right hand. Louis Pasteur showed that, as if by magic, a large number of ALL the left OR right handed molecules could get together and form a crystal containing just one or the other (although it is never possible to predict in advance if the crystal would be made up of all-left or all-right handed molecules). In origin-of-life terms, I think that an analogous transition occurred where molecules that could replicate or copy (along with the probability of being copied) could spontaneously occur.

What would this look like? In chemistry this phenomenon would manifest when a chemical soup makes the move from passive 'dead' molecules to self-replicating molecules, beating a path towards evolution. In artificial intelligence it is even simpler - it is the moment at which the computer system starts to generate interesting new things, and you don't want to turn it off.

Will we understand the origin of life, build artificial living machines with powerful artificial intelligence? Yes of course, I have no doubt about that, and we need to think about and debate the consequences. But it is probably much further away than many think. In the meantime we are embarking on a series of projects to explore chemical space, to see if we can engineer a system that displays 'Spontaneous Novelty'. We are also developing a theory to aid us in our search. The hope is that the theory will help us come up with experiments that will lead to new understanding and ultimately, the synthesis of new life forms. Should we be worried? I don't think so, but we should debate the merits of such research. Even if we are successful, what we are doing will produce systems many millions of times simpler than an Amoeba and much more fragile. Cyberdyne and the Terminator aren't coming any time soon.