WE HUMANS are in trouble. We have let loose a new evolutionary process that we don鈥檛 understand and can鈥檛 control.
The latest leaps forward in artificial intelligence, with its large language models and deepfakes, are rightly causing anxiety. Yet people are responding as though AI is just one more scary new technology, like electricity or cars once were. We invented it, the argument goes, so we should be able to regulate and manage it for our own benefit. Not so. I believe that this situation is new, serious and potentially dangerous.
Advertisement
My thinking starts from the premise that all design anywhere in the universe is created by the evolutionary algorithm. This is the simple, three-step process in which some kind of information is copied many times, the copies vary slightly and only some are selected to be copied again. The information is called the replicator, and our most familiar example is the gene.
But genes aren鈥檛 the only replicator, as Richard Dawkins stressed in The Selfish Gene. People copy habits, stories, words, technologies and songs; we change, manipulate, recombine and pass them on in ever greater variety. This second replicator, evolving much faster than genes ever could, Dawkins called memes 鈥� and they are selfish too.
Once I had fully grasped the idea of selfish memes, I realised how profoundly it changes our notions of human minds and cultures. I wrote The Meme Machine in response. In this new view, we are the meme machines, and memes compete to use us for their own propagation, creating not just silly videos and maddening ads, but all of our rich, evolving cultures. Of course, we humans try to select only those memes that make us happy or healthy, but the memes don鈥檛 care because they can鈥檛 care.
As we face up to the recent explosion in AI, new questions arise that both fascinate and worry me. Could a third replicator piggyback on the first two? And what would happen if it did?
For billions of years, all of Earth鈥檚 organisms were gene machines, until, about 2 million years ago, just one species 鈥� our ancestors 鈥� started imitating sounds, gestures and ways of processing food or making fire. This might not seem momentous, but it was. They had inadvertently let loose a second replicator and turned us into meme machines, with dramatic effects on the rest of life on Earth. Following the same principle, could a third replicator emerge if some object we made started copying, varying and selecting a new kind of information?
It could, and I believe it has. Our digital technology can copy, store and propagate vast amounts of information with near-perfect accuracy. While we had mostly been the ones selecting what to copy and share, that is changing now mindless algorithms choose which ads we see and which news stories they 鈥渢hink鈥� we would like. AIs equipped with language processing and 鈥減ersonalities鈥� become ever harder to distinguish from real people as they compete with each other and with us to get our attention 鈥� and the successful ones aren鈥檛 necessarily the most benign. Even though we built this amazing digital world for our own purposes, once a digital replicator takes off, its products will inevitably evolve for its own benefit, not ours.
All is not lost, though. We already cope with fast-evolving parasites such as viruses by using our immune systems, medicines and vaccines. Now, we need to build our collective mental immunity, our critical thinking and our ability to protect our attention from all that selfish information that tries to exploit us. Taking lessons from evolution, we can stop imagining we are the controllers of our accidentally dangerous offspring and start learning how to live with them.
Susan Blackmore is the聽author of The Meme Machine and many books on consciousness. She is聽visiting professor at the University of Plymouth, UK
