A self-replicating form of artificial life has arisen from a digital 鈥減rimordial soup鈥� of random data, despite a lack of explicit rules or goals to encourage such behaviour. Researchers believe it is possible that more sophisticated versions of the experiment could yield more advanced digital organisms, and if they did, the findings could shed light on the mechanisms behind the emergence of biological life on Earth.
While the process of evolution is well understood, little is known about how inert molecules first came together to form life. To investigate how humble beginnings can lead to complex ends, at Google and his colleagues designed experiments where tens of thousands of separate pieces of computer code randomly mingled, combined and executed their instructions over millions of generations.
Advertisement
Because there were no rules to govern how the code samples should change and no rewards for certain behaviour, the researchers expected the population, which was capped at a fixed number, to remain random and do nothing coherent. But to their surprise, they found that the simulation eventually led to the emergence of self-replicating programs that quickly multiplied to hit the population cap. Eventually, new types of replicators emerged that competed for space and occasionally overwhelmed and replaced the previous population, just as biological organisms can outcompete each other.
This research is far from the first attempt at mimicking life digitally: for example, simulations such as the Game of Life, which has a grid of cells that are either 鈥渁live鈥� or 鈥渄ead鈥� and are governed by simple rules, have shown self-replicating behaviour. Laurie says what makes this work unique is that the system had no formal rules, goals or processes to encourage or kick-start artificial life 鈥� it simply arrived. 鈥淚t all fizzes around and then suddenly: boom, they鈥檙e all the same,鈥� he says.
The experiments may not tell us anything concrete about how biological life began, says Laurie, but they do reveal that there are inherent mechanisms to create complexity from nothing. He believes that complex biological life is merely a result of a similarly long period of random iteration. 鈥淚 don鈥檛 think anything magic happened,鈥� he says. 鈥淧hysics occurred, and it just occurred a lot over a very long time, and it gave rise to some very complicated things.鈥�
But life on Earth emerged only after 鈥渂illions of years of massively parallel experimentation鈥�, says Laurie, and while he believes that great complexity would emerge from the team鈥檚 system if scaled-up in size and duration, we would quickly hit the limits of what can be performed with current computers.
鈥淢y gut feeling is that if you want more interesting behaviour, if you want things eating each other, or war going on between different species, or complexification that allows sensing of the environment and things like that 鈥� which I think would ultimately come 鈥� it鈥檚 going to require so much compute that we鈥檙e not going to practically do it,鈥� says Laurie.
Indeed, many of the team鈥檚 experiments ran for millions of steps before displaying organised behaviour. Laurie says that one instance, running on his laptop, involved processing about 3 billion instructions a second and it still took around half an hour for self-replication to emerge.
at the University of York, UK, says that the work is fascinating. 鈥淢anaging to evolve self-replicating programs from random starting points is a great achievement,鈥� she says. 鈥淭his is definitely a great step towards understanding potential routes to the origin of life, here in a medium quite removed from the standard 鈥榳etware鈥� of biology.鈥�
at the University of Southampton, UK, says the researchers鈥� results are 鈥渧ery cool鈥�, but notes they are unlikely to lead to increasingly complex behaviour automatically.
鈥淭he complexity, as they measure it, goes up after the onset of the self-replicator. But it鈥檚 not clear that it 鈥榯akes off鈥� in an interesting way,鈥� he says. 鈥淪elf-replication is important, but it would be a mistake to believe it鈥檚 a magic bullet from which everything else that鈥檚 exciting about life follows automatically鈥�.
at University College London is also doubtful that the work sheds light on the origin of life on Earth. She likens it to a classic experiment where RNA strands were replicated in a test tube, resulting in the length of the RNA getting shorter and shorter and replication speed getting faster and faster. A very simple form of natural selection was rewarding a lack of complexity, not encouraging greater complexity 鈥� the opposite of what is needed to explain the origin of complex life, she says.
鈥淗aving infinite copies of something does not guarantee complexity,鈥� says Nunes Palmeira. 鈥淚f you have one thing that just self-replicates and does it faster than everything else, then you鈥檙e just going to have a system that is completely taken over by that.鈥� By contrast, life involves multiple interacting components, including DNA, RNA, proteins and more, she says. 鈥淚t鈥檚 a very complex system, and I think we鈥檙e not that much closer to understanding how it arose de novo by just looking at self replication.鈥�
arXiv
