OUR feelings about a new gadget follow a familiar cycle: there is excitement at a fresh toy, then confusion as we struggle with how it works, then we get bored and chuck it in a cupboard. That, roughly, is what protagonist Charlie does in Ian McEwan鈥檚 new book (Jonathan Cape). But the gadget is an Adam, one of the first humanoid robots and fruit of a world in which Alan Turing stayed alive. Set in an alternative 1980s, where a weakened Conservative PM (Margaret Thatcher) battles a socialist leader (Tony Benn) over the UK鈥檚 future in the European Union, McEwan tackles Turing鈥檚 big question: can machines think?
You鈥檝e often written about science and scientists, but this is your first science fiction 鈥�
I don鈥檛 really see it as science fiction. I鈥檝e never been much interested in the remote future. Especially when people are crossing the galaxy at five times the speed of light and wearing anti-gravity boots. I could never suspend my disbelief. I come at this more in a personal way. What it would be like to have, in front of you, in your house, in your life, a machine that tells you it can think, and poses the problem for you as to whether you believe it.
Advertisement
鈥淭he extent to which we devolve moral decisions to machines is going to be a very interesting ride鈥�
With recent real developments in artificial intelligence, did you feel it was time to write this?
It was. I had a long chat with Demis Hassabis [head of Google鈥檚 sister AI division DeepMind]. He鈥檚 really become one of the rock stars of AI. In a way, it鈥檚 like all my novels in that I don鈥檛 think I鈥檓 doing research. At the time, it鈥檚 just interesting me, and then gets to a point where suddenly it鈥檚 the next thing for me.
Do you have an Alexa or a Siri?
We have an Alexa.
How do you converse with it?
Mostly irritably. It鈥檚 kind of worn off for me. I think there are these technologies, which I mention at the beginning of the novel, that come and everyone鈥檚 excited about them, then we realise they don鈥檛 actually fit the human pattern. So the fridges that could tell you when your eggs are rotten are completely extraneous. But I鈥檓 plugged into all the usual stuff that we can no longer live without.
I realised last night, when we left to come here, suddenly the internet interface with our central heating didn鈥檛 work. Just two years ago it would never have occurred to me that you could turn on your heating remotely. And now we would just consider it a real nuisance, like as if the pipes had burst or something. The speed with which we absorb these things and either discard them or find them boring when they break down is quite astonishing.
Do you think we鈥檙e in trouble because we have become so reliant on these technologies?
With AI, we鈥檙e going to have that in spades. Already we鈥檙e having to think ethically about autonomous vehicles, and what kind of choices they鈥檙e going to make. Do they run over the granny, the dog, the child, or allow the 鈥渄river鈥� just to kill themselves in a head-on crash?
We鈥檙e suddenly having to devolve these choices to someone else, to something else. The extent to which we devolve moral decisions to machines is going to be a very awkward and interesting ride. I鈥檓 sorry to be 70 and not see more of the story. The area where our interaction with machines enters the moral domain is going to be a field day for novelists.
[video_player id=鈥漍oralOsM鈥� access_level=鈥漞veryone鈥漖
Your 1980s has more advanced tech than we do 鈥� because Alan Turing flourished. Do you think he would have helped create it?
Who knows? Turing was young when he died and would have lived through the 60s, and as a gay man maybe would have benefited from a loosening of the terrible laws that condemned him. And he may or may not have had an impact on the computer revolution. He might have gone off into biology. I have no idea.
For the purpose of this novel, I have reinvented the past. There鈥檚 something completely arbitrary and contingent about where we are in the history of science. It could have been earlier, it could have been much later. And it could have been elsewhere, too. So let鈥檚 push the science on, and let鈥檚 have Turing not commit suicide, then why not shift everything so that the politics is different, but not entirely different?
There鈥檚 a very Brexit feel to the novel. Did you start out wanting to marry it with AI morality?
I didn鈥檛 even think about it. I just live in it so much. I鈥檓 a Brexit junkie, so I have to read at least 50,000 words a day on it. It makes me miserable, but I can鈥檛 leave it alone. So even if I decided to keep it out, it probably would have been in there anyway since it鈥檚 infused my every last neuron.
When you designed Adam, was that something you did as a whole or as the story went along?
I take a novelist鈥檚 licence. He has a body. He has sex. He has mucous membranes. All these things are, who knows, 200 years away. But I needed all that in order to get to the matter of how we鈥檙e going to deal with what it would be like. I was fantasising about the point where it would become extremely rude to say to you, 鈥渁re you real?鈥�.
At first, people would think it鈥檚 very PC not to ask. Then, actually, we just wouldn鈥檛 know. And then we wouldn鈥檛 know whether the prime minster was real. So I played with the idea that it鈥檚 like the Turing test taken to extremes. If you can鈥檛 tell, then you just have to assume you must treat them exactly as if they were human.
One of the highlights of the novel is when Adam becomes extremely attached to generating poetry. Have you experimented with this?
No, I have no interest in AI-generated poetry. Of course, I speak here as a novelist. When a robot could write a novel that you could not tell was written by a robot, in which it had the whole range of human interactions, psychological insight, understanding of the subjective nature of consciousness, then I鈥檇 say we were just about there. At that point, I would have to say you may now sit on juries, if you can write a good novel.
Would you welcome that?
I guess it would be like, dare I say it, a bit like when we had that argument about the Booker prize in which we let Americans come in. If we get to the point where we say, well, a human hasn鈥檛 won the Booker prize for 50 years, we say, yes, but look at what wonderful books we鈥檝e been reading. Or a robot might be saying that to us.
You seem intensely relaxed about the prospect.
I鈥檒l be dead. And I don鈥檛 really hold out much hope for an afterlife, so I won鈥檛 be watching. Who knows? The whole thing might be irrelevant because we鈥檒l have a nuclear war in the next 15 years.
Does that concern you? Should we worry about nuclear war or climate change rather than AI?
I never like to say what people should worry about. But yes, I think there鈥檚 a real possibility that we might not get through the 21st century without a nuclear war somewhere. India, Pakistan, China versus the rest. Russia, Europe, America. Or Saudi Arabia and Iran. There鈥檚 plenty of possibility.
Is that a problem that AI could solve, do you think?
It might be a very logical decision of AI to, say, launch a pre-emptive attack. Or think that if nine-tenths of humans were eradicated that it might be a kind of victory. Who knows what logic applied to this would do? In fact, I have Adam speculate, if you wanted to cure cancer then you鈥檇 just have to kill all the humans. And then no one would have cancer. You have to watch where logic takes you in this.
