ONE question to ask of any interview in the papers is “What are they selling”? Although the answer is seldom as blatant as with vapid celebrity interviews that always end with the statement that “Grizelda Foxbunny is a brand ambassador for Labradoodle Shampoo,” it is vanishingly rare that any discussion of cultural matters is not pegged on a book or a film to sell.
The only reliable exceptions to the rule are the LRB’s reviews, which can be anything up to two years late, and the FT’s “Lunch with . . .” Saturday interviews, which are run whenever interestingly busy people find the time to talk.
This week, it was Geoffrey Hinton, who won a Nobel Prize for his contribution to the mathematics of artificial intelligence. He has no doubt that the machines are intelligent: “It seems very obvious to me. If you talk to these things and ask them questions, it understands,” he says.
Of course, he doesn’t have to believe the answers: “[ChatGPT] featured in his recent break-up with his partner of several years. ‘She got ChatGPT to tell me what a rat I was,’ he says, admitting the move surprised him. ‘She got the chatbot to explain how awful my behaviour was and gave it to me. I didn’t think I had been a rat, so it didn’t make me feel too bad. . . I met somebody I liked more: you know how it goes.’”
None the less, he believes that this form of intelligence will outstrip human capacity and control within 20 years: “Hinton believes ‘the only hope’ for humanity is engineering AI to become mothers to us. . . There is only one example we know of a much more intelligent being controlled by a much less intelligent being, and that is a mother and baby. . . If babies couldn’t control their mothers, they would die. . . ‘That’s the kind of relationship we should be aiming for.’”
What seems odd is his assumption that that this kind of care and compassion could be programmed into a machine at all — and that, if it were, it would be universally adopted. He is quite clear about the dangerous motives of the companies presently leading the AI boom: “‘What’s actually going to happen is rich people are going to use AI to replace workers,’ he says. ‘It will make a few people much richer and most people poorer. That’s not AI’s fault, that is the capitalist system.’”
THIS made an interesting contrast with a long blog post by Mustafa Suleyman, one of the founders of Deep Mind and now at Microsoft. He is worried that people will take AI for conscious, because, he says, the computing power and the programming techniques that we now have mean that, within two years, there will be AIs that “imitate consciousness in such a convincing way that it would be indistinguishable from a claim that you or I might make to one another about our own consciousness. . . A wide variety of people will be able to create something like this. As such . . . it will be relatively easy to reproduce and therefore very widely distributed.”
What he proposes as a response, or as a preventative measure, is the creation of an AI that does not claim to have “experiences, feelings or emotions like shame, guilt, jealousy, desire to compete, and so on. It must not trigger human empathy circuits by claiming it suffers or that it wishes to live autonomously, beyond us.”
This comes too late. Existing AIs will already do everything that he wants forbidden if they are prompted correctly. Obviously, it would be possible to produce AIs that would recoil in horror at such indecent suggestions, or at least respond primly that they were not that sort at all; but the market would, I think, prefer machines that were more fun. If we are in the middle of a loneliness pandemic, many hundreds of thousands of people are going to want an imaginary friend in their phones, and many billions of stock market dollars have been bet on the proposition that they will pay for it.
This is not quite the future that Hinton fears: the machines will not control people for their own purposes: they will make it easier for our own desires to control us to our disadvantage. To fight that demands a recognition that we are fallen creatures, and a willingness to admit that sometimes there is no health in us. Forward-looking churches had better prepare 12-step programmes for people whose lives have been ruined by their interactions with a chatbot.
