Ian McEwan has no interest in science fiction. That’s why, when the author decided to write a novel about a messy love triangle between two humans and an artificially-intelligent android, he opted to reimagine the past instead of predicting the future. How do you rationalise an 80s London where synthetic humans accompany their flesh-and-blood owners to the corner shop? By bringing Alan Turing back to life.
“You only have to look back one or two steps to see how all the present could easily have been something else,” McEwan says. In Machines Like Me, he imagines that the mathematician and codebreaker didn’t die by suicide at age 41 but lived on to crack P versus NP, a major computer science problem that in reality remains unsolved.
In the world of the novel, this opens a path toward artificial minds that can learn, think – and fall in love. “A long interest I've had for years now is what it would mean to have artificial intelligence,” McEwan says. “What would happen if that was extended to the notion of not only general intelligence but into the emotions, and how would we respond to an intelligence as great or greater than our own?”
To help him grasp the intricacies of AI, McEwan turned to Demis Hassabis, the neuroscientist and co-founder of Google-owned AI company DeepMind. Over dinner at a London steakhouse, Hassabis explained how he’d built the computer program that in 2016 defeated Lee Sedol, the 18-time world champion of the Chinese board game Go, by training it on data from thousands of amateur and professional games. After the meeting, McEwan ended up inserting Hassabis into his novel, making him one of Turing’s collaborators. “I wrote to Demis and said I was thinking of putting [him] in a novel,” he says. “I didn't get a reply, so I thought, ‘Put him in anyway.’ He comes out all right.”
In the novel, the synthetic human – one of a batch of experimental androids called Adams and Eves – finds itself caught in a moral quandary. It has fallen in love with a woman who, the android suspects, may have committed a crime in order to avenge the death of a friend. A human might be willing to forgive a crime in such circumstances, but how would machines navigate moral grey areas? Humans are better at creating moral rules than we are at following them, McEwan says. Machines with morals might be morally better people than we could ever be.
That needn’t be such a bad thing. “Most of us are surrounded by people who are more intelligent than us,” McEwan says. “There's always someone who can run faster than you, is better looking than you, or has more capabilities intellectually than you. I'm not sure we'll have that much of a problem with that. I think we could fall in love with these machines.”
That’s if they stick around long enough for us to learn to love them. It doesn’t take long for the latest piece of technology to go the way of the fondue set or mouse mat, McEwan says, and predicting what the next piece of revolutionary technology is – or how it will end up being used – is a fool’s game.
“It's like the seven billion of us are sitting around a vast Ouija board and we're all moving, and it's writing sentences that none of us could predict,” McEwan says. “[The future] is in our hands and yet it's constantly a shock and sometimes, maybe less frequently, a pleasant surprise.” Matt Reynolds
Machines Like Me is published by Vintage and will be released on April 18, 2019
This article was originally published by WIRED UK
