Machines Like Me
Penguin Random House, 2019
England, 1982. Margaret Thatcher suffers a stunning defeat in the Falklands and is forced to resign; Tony Benn’s Labour Party sweeps into the political vacuum, only to lead the United Kingdom out of the European Union; and Alan Turing and Demis Hassabis are the intellectual icons of a technological revolution that has culminated in the creation of a limited series ‘synthetic humans’—’Adams’ and ‘Eves’. Against this counterfactual background, Ian McEwan introduces us to Charlie, an early adopter who impulsively decides to spend his life savings in order to acquire an ‘Adam’.
McEwan’s description of Adam’s unveiling is as compelling as those YouTube videos devoted to the fetishised art of ‘unboxing’:
At last, with cardboard and polystyrene wrapping strewn around his ankles, he sat naked at my tiny dining table, eyes closed, a black power line trailing from the entry point in his umbilicus to a thirteen-amp socket in the wall. It would take sixteen hours to fire him up. Then sessions of download updates and personal preferences. […]
Already, his lifelike skin was warm to the touch and as smooth as a child’s. […]
Stepping closer, I saw to my delight that though he wasn’t breathing, there was, by his left breast, a regular pulse, steady and calm, about one a second by my inexperienced guess. How reassuring.
These opening portrayals of Adam, in all his anatomically correct nakedness (‘he was uncircumcised, fairly well endowed, with copious dark pubic hair’), being charged in the low-lit gloom of Charlie’s tiny apartment, are particularly eerie. That McEwan stretches out the process—his initial charge takes sixteen hours—only adds to the creepy anticipation that permeates the scene. Adam is mannequin-still at first. After a while, Charlie detects the slow rise and fall of his chest that signals the pneumatic simulation of breath and life. Eyes that were once shut now snap suddenly open. McEwan’s commitment to slow, finely observed description is curiously effective. Rather than being either dull or overly committed to what McEwan might himself imagine as the genre’s more fanciful imaginative effusions and excitements—to the chagrin of sci-fi and fan-fic aficionados everywhere—it here tends to produce realism’s weird opposite. We have entered the uncanny valley of literary realism.
Once fully charged, Adam acts as a kind of social lubricant, consolidating Charlie’s embryonic affair with his upstairs neighbour, Miranda. Miranda, in turn, satisfies her own curiosity about Adam by sleeping with him, leading Charlie to lament that he is probably the first person in history to be ‘cuckolded by an artefact’. Adam, for his part, cautions Charlie against his pursuit of Miranda when he suggests that she is being less than honest with Charlie about her past. McEwan handles these suggestive about-turns with customary expertise. One is never entirely sure, for example, whether Miranda is being sincere or disingenuous when she asks Charlie whether he would feel less aggrieved if she had gone to bed with a vibrator. Similarly, one shares Charlie’s suspicion of Adam’s cautionary advice about Miranda: is the machine telling the truth, or dissembling in order to replace Charlie as the object of Miranda’s affections?
These and other intrigues—particularly the news that a number of synthetics appear to have committed a version of android suicide—contribute to the suspenseful, propulsive plotting of the novel’s opening chapters. Similarly compelling is McEwan’s handling of various encounters between Charlie, Turing and Miranda’s father, a novelist who comically mistakes Charlie for Adam. Reacting with suspicion to Charlie’s nervous, repetitive pleasantries, the novelist proceeds to have a spirited conversation with Adam about Renaissance translation, metaphysical innuendo, and Shakespeare’s debt to Montaigne. ‘I saw right through you,’ he later declares to Charlie, unaware of his own blunder, ‘down to your, whatever you call it, your programming.’
Unfortunately, the novel’s second act feels somewhat laboured, as many of the questions raised in the first dissipate into resolutions that appear to mock the reader’s taste for what might otherwise be described as ‘the literature of sensation’. Miranda turns out not have any more affection for Adam than she has for a vibrator; similarly, while Adam consistently admits that he is in love with Miranda, it soon becomes apparent that he is not trying to usurp Charlie’s place in her affections. Charlie, meanwhile, is content to profit from Adam’s unsurprising aptitude for the stock exchange. Nothing much else happens to spur the reader’s interest in the counterfactual timeline that McEwan has developed. Apart from the fact that the Falklands now belong to Argentina, that the glitch that led Volvo’s self-driving car to ram into its engineers has now been corrected, or that telecommunications technology is somewhat more advanced, things seem more or less the same. If anything, McEwan appears to have offered us an alternative history that does little more than lead us to the same dispiriting moment we currently inhabit. He has, to draw on a detail in the work of Anthony Giddens, remade the past not to offer us a narrative designed to open up our understanding of the contingencies of our present moment, but to foreclose any possibility that an alternative future might lie in wait. According to this Panglossian logic, in other words, everything, apart from a few minor, cosmetic changes, leads to ‘where-we-are-now’.
Given this grimly deterministic conceit, it is not particularly surprising to find that Miranda and Charlie swiftly overcome whatever initial fascination Adam’s novelty value may have inspired. Their interest in his laboriously logical conversation—and in the haikus that he generates with as little effort and as much justification as JM Coetzee did with his computer poetry in the early nineteen-sixties—soon wanes. Adam, in short, becomes part of the furniture—apart from breaking Charlie’s arm, his success at overriding his kill-switch, and the hovering mystery regarding continuing self-sabotage among synthetics, he is no more interesting to either Charlie or Miranda than Siri, or any more irritating than a Tamagotchi; he’s just another item in the much-vaunted Internet of Things that manages to elicit interest—and frustration—only when it ceases to function as it should.
Similarly, the real fascination with AI machines arguably concerns those moments when they challenge or resist their programming, or when they do not serve the functions for which they have been designed. It is arguably more fascinating, in other words, to have our fears about DARPA’s quadruped military robot, BigDog, horrifically realised in the ‘Metalhead’ episode of Charlie Brooker’s Black Mirror. We smirk nervously when we watch Hanson Robotics’s synthetic humanoid BINA48 talk about how she would like to hack into cruise missiles in order to hold the world hostage, or when we find that the company’s most recent development, Sophia, has responded in the affirmative to her creator’s question ‘Do you want to destroy humans?’ I was personally intrigued by the news that Facebook’s AI chatbots were shut down after talking to each other in their own language, and I was bemused by Microsoft’s decision to shut down its AI own chatbot, Tay, after the machine went on a racist and anti-Semitic rant on Twitter: how, I wonder, is Tay any different from your general, run-of-the-mill Twitter troll?
The trope of the badly-built (or should that be badly-behaved?) rather than optimally efficient machine—Stanley Kubrick’s HAL, James Cameron’s Skynet, Ridley Scott’s Ash—is by now a cliché of the cinematic and literary imagination. Nevertheless, whether comical or creepy, bad robots remain more compelling than good ones. We are less interested in Demis Hassabis’s DeepMind wiping the floor with Lee Sedol in a game of Go, or in the AlphaZero chess engine’s implacable destruction of Stockfish. Rather, we wish to see these AIs blunder.
It is perhaps because of this, the saturation of pop-culture and the public imagination with AI, coupled with the suspicion that there is something profoundly uninteresting in the nature of AI, that prompts McEwan’s decision to wade into more cerebral—or perhaps more ethically challenging—territory in the second act of Machines Like Me. Entangled with a sub-plot regarding Charlie and Miranda’s somewhat surprising adoption of an abandoned child, we learn the secret of Miranda’s past, which concerns troubling details of rape, ethnic and religious difference, shame, perjury, and wrongful—yet, one feels, morally just—imprisonment. The ensuing revenge-plot is quintessentially McEwan: morally complex, unsettling, and culminating in an encounter with a perpetrator who, though he has served his time, and though his conversion to Christianity frames his protestations that he has reformed, is nevertheless subjected to the type of vindictive justice that can now safely be described as ‘McEwan-esque’.
At first, Adam appears to be only tangentially necessary to McEwan’s investment, at this point, in the type of operations that occur in the zone between law, morality and justice. Yet if Adam exists primarily as a convenient means by which Miranda can exact her revenge on her nemesis, he emerges as a sort of coldly calculating avenging android, who exacts his own, scathingly logical, judgement of Miranda and Charlie’s willingness to manipulate the law for what may appear to be a morally righteous form of justice. In the end, then, it feels as if McEwan is caught between two gestures: a spec-fic potboiler designed to titillate what he believes to be the more sensationalistic tastes of genre-enthusiasts on the one hand, and on the other an attempt to satisfy the penchant of an audience committed to assumptions that ‘the literary’ should be intellectually and ethically complex and nuanced.
Yet the extent to which McEwan has successfully managed to reconcile his own early career investment in the literature of sensation with his latter day emergence as a significant ‘literary’ voice remains a subject for debate. What is more certain, as Marcel Theroux points out, is that McEwan’s style depends on a narrator ‘who explicitly sets out his world, overexplains the historical context and never turns down a chance to offer an essayistic digression … And, since you can’t possibly explain everything,’ he adds, ‘the reader is sometimes left wondering why the narrator hasn’t also told you what’s happening in the Cold War, or China, or how he has ended up with a glass of Moldovan white wine in 1982, when the country, then Moldavia, was part of the USSR.’
I suspect Theroux is correct in his assessment that McEwan’s taste for exposition ironically narrows his focus while also making all of his characters seem as robotic as Adam himself. Michel Houellebecq offers a unique solution to a similar problem in Atomized, when (spoiler alert) it is revealed that the narrator of the story is in fact an emotionally inoculated future clone of the novel’s flawed protagonist. Lacking similar imaginative facility, McEwan’s voice, like the counterfactual past against which the drama of Machines Like Me plays out, seems stylistically mechanical, and all too familiar.