At a time when humans are still struggling to define intelligence in machines, an exhibition about the same could be heading into controversial territory.
So they’ve avoided the question at the Barbican, preferring to put on a display of the instruments of intelligence and leaving any existential questions alone.
An interesting opener to an investigation into the future of intelligence looks deep into the past though, to a time when inanimate objects were thought to be able to have souls and intelligence of their own.
It gives visitors a moments pause to wonder how much of non-human intelligence will be the result of the machines being clever, and how much will be us anthropomorphizing the machines.
The rest of the curved gallery is given over to objects that both stand alone as objects to see, and part of a longer narative through the glowing and noisy gallery space.
It’s oddly then a comprehensive storyline, but also delivered in isolated items, which gives the impression of a display where they’ve been offered objects, and then tried to create an exhibition from them.
There are a number of moments in computer history here though.
Alan Turing’s ghost lingers with letters and an actual Enigma machine on display, and his impressions of the practical difficulties of Artificial Intelligence.
Deep Blue, a computer that became famous not for processing paperwork faster than a human, but for calculating chess moves faster than a human. Its iconic status is odd — would we be as excited about a computer processing your payroll slips faster than an office of humans?
Here on display is the brain.
Will future machines look on such a display of a computer sliced up and shown off in the way that we might today balk at the sight of a human brain sliced for show?
The heart of the exhibition is given over to the current mass wave of machine learning – the ability to process vast amounts of information to find patterns and “learn” about them.
It’s a fundamental question as to how much of what we call AI, is not intelligence, but the ability for pattern recognition. If all an AI computer can do today is process information very fast, then is it really intelligence? Would a million people working collectively on a problem using abacus for their tool be thought to be just calculating a problem, or would their collective effort be thought to be in itself intelligent?
I see rather too many companies claiming now to be using AI to solve problems that are often not being solved by anything more sophisticated than a modest computer programme. Will AI be the “digital” of the marketing world, where in the 1980s, almost anything was seen as better if it the word “digital” in the brandname, even if in reality the digital element was at best, negligible.
One question which is raised is how much of AI’s future is replicating human biases. There’s a large section of the exhibition that highlights how difficult it is for facial recognition software to recognise black faces, and elsewhere people worry about unexpected racial biases cropping up in patterns.
At the end is one of the creepiest visions of the future though.
A human like android on a metal skeleton waving hands around and making the most unearthly moaning sounds. A machine is dying, and as we watch the mouth attempting speech and staring at us, we wonder again how much of non-human intelligence will be the result of the machines being clever, and how much will be us anthropomorphizing the machines.
Entry is £17 at weekends and £15 during the week, with late openings on Thur and Fri to 9pm.