Ken Gordon | Essays

Designers Like You Should Read Machines Like Me

People, you might have noticed, are wracking their brains to understand artificial intelligence. To assist them, authors such as Eric Topol, Andrew McAfee and Erik Brynjolffson, and John C. Havens have published good, clear books, for a general audience, on AI. But I recently discovered something: these volumes, while all ably written, lack something.

I didn’t realize the missing element until I encountered a new book. A work of fiction. Machines Like Me is a novel by the best-selling, prize-winning Ian McEwan. What does McEwan have that the other authors don’t? A world-class, world-generating imagination that can envision, at book length, a realistic robotic future. Machines Like Me isn’t just about a world with robots: it is a world with robots.

A book like Machines Like Me is necessary, if we want to design the artificially intelligent future in an ethical manner. In The Ethics of Reading, J. Hillis Miller writes: “the moral law gives rise by an intrinsic necessity to storytelling, even if that storytelling puts in question or subverts the moral law. Ethics and storytelling cannot be kept separate, though their relation is neither symmetrical nor harmonious.” If we’re serious about creating an ethical code or guidance around AI, we can’t conduct the investigation on a purely abstract plane.

Machines Like Me is concerned with the ethics of robotics—Do robots have rights? What might they be? Can a being with artificial intelligence be considered a person?—but not in some detached manner. The book is narrated by a self-absorbed, 32-year-old Londoner name Charlie Friend. His mother dies, he inherits money, he spends £86,000 on a synthetic human. Adam. (In this book, all the robots are named, biblically, either Adam or Eve.) Charlie is a good person to tell this story: he studied anthropology at university, loved electronics as a kid, and wrote a short book on artificial intelligence—though, weirdly, was never professionally involved with AI again. Most importantly, Charlie loves Miranda, the beautiful young woman who lives next door with a secret. It is into this thick dramatic context that McEwan places Adam.

Non-fiction writers are bound by the idea of fact. They might make brief speculative ventures into the future—Havens does this in Heartificial Intelligencebut these scenes aren’t very long or executed with McEwan’s Man Booker Prize-winning finesse. In Deep Medicine: How Artificial Intelligence Can Make Healthcare Human Again, Eric Topol, perhaps the world’s leading authority on digital health, imagines several conversations between patients and their virtual assistants. In one, a patient named John and his virtual assistant, Ann, have the following exchange:
“John, I got your oxygen saturation last night, dipping down to 67.”

“Ann, I forgot to put on my BiPAP mask.”

“John, your blood pressure went up to 195 at the time, and it was high all night long, mean of 155 systolic.”

“So not just my sleep apnea, Ann?”

“No, John, your weight gain of 12 pounds and no exercise in the past week is probably contributing.”

“That’s because my back was out of whack and I sat around eating.”

“Yes, I have been warning you about that!”

“OK, Ann. That’s enough. I quit. You’re fired.”
Yes, this passage gives us a sense of the future of conversational computing. Unfortunately, it also reminds us Topol’s limitations as a writer of dialogue (too many lines, for instance, begin with “John” or “Ann” or “No, John” or “OK, Ann” and John’s kiss off simply lacks conviction). We need something more from such scenarios, if we’re to create, or be properly prepared for, a positive robotic future. Machines Like Me provides it.

The novelist gives artificial intelligence a human voice (pardon the paradox). Think about it: Novelists fabricate synthetic characters that are so life-like, we believe them when they speak. That skill makes them the ideal people to articulate an android’s point of view. Where exactly will a robot fit into the social scheme of human life? Is it a friend? relative? employee? Very difficult to say. And one can imagine—McEwan does!—how tricky this will be with the uncanny intellect of someone like Adam.

Consider a scene from early on in Adam’s life of sentience, Charlie takes him around to his newsagent, Simon Syed. After a bit of small talk, Simon says that the anthropology journal he special-ordered for Charlie has arrived and is waiting for him on the top shelf. “You can get it yourself?” jokes Simon, who, taller than the narrator, typically grabbed the journal. Adam’s retort? “Your self, you say. There’s a coincidence. I’ve been giving some thought lately to the mystery of the self. Some say it’s an organic element or process embedded in neural structures. Others insist that it’s an illusion, a by-product of our narrative tendencies.”

In Machines Like Me, a robot joins what Freud dubbed the “family romance,” and things get deep and dark as the novel progresses. Difficult to imagine a non-fiction book imaging in any compelling way how sex and love might play out with artificially intelligent beings—but that’s precisely what Machines Like Me does. One night, after a fight, Miranda goes upstairs to her apartment and has sex with Adam (while Charlies listens!). “I could have run up the stairs and prevented them, burst into the bedroom like the clownish husband in an old seaside postcard,” he says. “But my situation had a thrilling aspect, not only of subterfuge and discovery, but of originality of modern precedence of being the first to be cuckolded by an artefact.”

Crazy, right? It was a one-time event, and Charlie and Miranda soon make up. Adam, however, spends the rest of the novel in platonic love with Miranda, writing her astronomical piles of haiku and saying in plain terms, to Charlie and Miranda, that he loves her. Late in the novel, we learn how Adam repeatedly asked Miranda to have sex again and, in an extremely pathetic moment, begged to masturbate in front of her. This is weird and horrifying… but there may well be weird and horrifying aspects of life with robots—and you’re not going to get them from a professorial review of the latest research.

The psychological dynamic of Robots Like Me is extremely complicated, in a rather Oedipal way. Let’s remember that Charlie and Miranda collaborate on Adam’s programming—it’s a method of co-parenting that will surely be part of our robotic future. Unlike a child, however, Adam can be shut down with a press of a button when he becomes ungovernable. One of the ways that the humans in Adam’s life control him, is via a kill switch, and Charlie and Miranda rush to press it when Adam gets in the way. The button-pressing goes on for a while, until one of the most startling scenes in the book.
I approached the table and as I passed behind Adam, I reached for the special place low on his neck. My knuckles brushed against his skin. As I positioned my forefinger, he turned in his chair and his right hand rose up to encircle my wrist. The grip was ferocious. As it grew tighter, I dropped to my knees and concentrated on denying him the satisfaction of the slightest murmur of pain, even when I heard something snap.

Adam heard it too and was instantly apologetic. He let go of me. “Charlie, I believe I’ve broken something. I really didn’t mean to. I’m truly sorry. Are you in a lot of pain? But please, I don’t want you or Miranda ever to touch that place again.”
The emotions in Machines Like Me can run high and deep, just as they surely will when we start designing robots, and bringing them into our homes. A novel like this reminds us that expanding the family of man to synthetic humans will surely be an emotional event.

The road of excess leads to the palace of wisdom... You never know what is enough until you know what is more than enough.” –William Blake

Machines Like Me is brimming over with social complexities. I haven’t even mentioned Miranda’s dying father, the little boy Miranda and Charlie are considering adopting, a rape, a lie, not to mention the fact that McEwan gave Alan Turing a happier—even heroic—biography, kept the Beatles intact, and reworked the facts of the Falkland Islands war. McEwan created, that is, a whole universe around Adam, and this universe, properly explored, could provide designers, ethicists, and entrepreneurs of the future will all kinds of ideas.

Seeing and feeling the implications of AI requires a leap of the imagination, not analysis. McEwan’s novel is much more than an argument about artificial intelligence, as most non-fiction AI works are. The nuances of the novel form give us a visceral feel how AI will change the way we live. It is, of course, just one writer’s vision… but surely it will inspire many designers, including you, to think deeper and better about our collective future. Machines Like Me shows us that bringing robots into our lives will bring challenges. As McEwan’s Turing says, robots like Adam were:
ill equipped to understand human decision-making, the way our principles are warped in the force field of our emotions, our peculiar biases, our self-delusion and all the other well-charted defects of our cognition. Soon these Adams and Eves were in despair. They couldn’t understand us, because we couldn’t understand ourselves. Their learning programs couldn’t accommodate us. If we didn’t know our own minds, how could we design theirs and expect them to be happy alongside us?

Posted in: Books, Technology

Jobs | July 21