Riveting questions given robotic answers AI: More Than Human Barbican, London EC2 ★★★★★
The rise of artificial intelligence, suggests the Barbican in the notes to its exhibition AI: More Than Human, is both “daunting and liberating”. How powerfully the show’s curators must have felt those twin sentiments as they plotted their exploration of a technology that, rightly, they estimate to be of central significance in our lives.
Liberating, because the concept of creating other thinking creatures has been fascinating us for so long that it is both intuitive and essential. It amply rewards exploration, and what riches there are to enjoy: Frankenstein’s monster takes its place among representations of Shinto deities and Western alchemical treatises and the pages of postwar comic books announcing the arrival of the golem, a creature rooted in Jewish folklore. This cheeky assembly of
the sacred and the profane nimbly conveys both our enduring temptation to wield for ourselves the spark of life, and also its faintly transgressive associations. With this conjuring, this magic, rite is not always right.
And so in its opening rooms, each a sinuous, curtained pod, the show successfully taps into a universal emotion. AI, we understand, may be a new world, but it is anchored in an ancient human desire to play God, to breathe life into beings in our own image, and then watch as they too transgress, no matter our avowed power to control them.
This first of four sections lays the foundation for what needs to be a piece of creative wizardry of its own. Because there is no escaping the daunting curatorial task to come: the sheer feat of charting the rise of AI over the past two centuries and then projecting its possibilities into the future. It would challenge the deepest neural network, the most powerful algorithm. Here, after all, is a highly complex technology. Evoking the universal creative urge must be exchanged for specific definitions – of machine learning, of deep learning. And here the show begins to falter.
How, in truth, could it not? For once you embark on explanation, there are vast amounts of explaining to do. Here, the exhibits are deeply involved: visitors model city infrastructure, or manipulate digital representations of brain cells. Stop and engage with each and your own brain will begin to ache.
Still, there is a thrill to the prime exhibits, even if some are replicas: Leibniz’s counting machine, Babbage and Lovelace’s Analytical Engine, Turing’s Bletchley Park Bombe, and an Enigma Machine – the evolution from calculation to computation laid bare. But the next step, from computation to AI – the spark of digital life – is altogether harder to evoke.
Anthropomorphism helps. The render of Marvin Minsky’s Tentacle Arm, with its steely humanity, provides a jolt of emotional discombobulation, as does Atilla, a spider robot from the Nineties.
But the truly unsettling aspects of AI today are much harder to get across. Bravely, the show tries anyway, tackling burning issues such as surveillance and algorithmic bias. But each requires more screens, more concentration. We are forced to learn, not feel, the points being made.
There are exceptions, such as Deepmind’s Alphago Zero algorithm beating the world’s best player of Go, which seemed to reveal a digital intuition at once different, and better, than our own. Its human opponent also explains how his triumph in the fourth game was one of his most satisfying. Suddenly, a novel relationship between AI and man emerges.
It is not until you face Alter 3, an android with a human face, whose eyes occasionally lock on your own, that you are made to confront the implications of the current technological revolution. Here is the uncanny valley – that creepy chasm between the almost human and the truly human. Will that valley be crossed? And if so, will we stop feeling creeped out as the golems of tomorrow look and feel like us? Will we in fact choose to blur the line between man and machine, which we defend so vigorously today?
This exhibition poses all the right questions. But the answers, largely, are conceptual. The challenge in communicating those concepts is to balance the didacticism of explanation with the artistry of evocation. Ultimately, it proves an impossible, if noble task.