AI: more than human?
A day out at the Barbican
By Laura Caccia
It was team day for Oxford Insights. Usually that means an American Breakfast (thank you Sabrina) a British presentation (thank you Richard) and some very Australian activity organisation (thank you Scarlet and Emma). It also means a mystery team building event.
Surprise! This time we found ourselves in the grand towering placidity of the Barbican, queuing up for the “AI: More than Human” exhibition. As AI storytellers and policy makers, we had been curious about this exhibition for months. Which out of the million faces of AI would they present? What story would they tell? Surely an AI exhibition can fill to fit any space; the vast maze of the Barbican felt very fitting.
Yet as we wound our way through the narrow dark corridor of the exhibition space, the maze seemed more intellectual than physical. They had tried to show all the most popular faces, and we couldn’t see where we were going.
There were a number of scattered and underdeveloped popular stories throughout the exhibition: spirited humanoid characters in popular culture, AlphaGo beating the world Go champion and AI in medical sciences. The exhibition islands were numerous, as were the flashing lights, videos and interactive screens. Through all this we kept searching for a story or a message, but found little other than a repetitive thudding beat to the tune of “AI is sexy and you’ll never understand”.
Connie (pictured above): “I just want a still image!!!”
When we did manage to focus there was an interesting spread of AI-related stories, including how Massive Attack have coded their album “Mezzanine” into a strand of DNA and filled a spray can with album coded paint. We also saw some beautiful early computers, letters to and from Alan Turing, and some fascinating footage about algorithmic bias.
However, while it’s important to showcase the achievements of the scientists in the past few centuries who have brought us to this point, the result of presenting centuries of “what” without the “how” is alienating, to say the least. The basic concepts of AI are not difficult to understand, but we learnt very little about the basic concepts in this space.
As a society we need to break what is essentially a powerful algorithm out of the mythological aura of, as CJ puts it, the “Frankenstein-esque behemoth about to destroy humanity” that it has caught like the flu. This “very 2000s conception of artificial intelligence” ignores the fact that now “we actually have a much more sophisticated understanding of AI”. This is an old story. Only when we know what AI can and cannot do can we figure out its impact on our future, and what benefits or risks it might pose. Understanding the “how”, in this case, is key.
Obviously this is far too boring to sell exhibition tickets, so on we go.
Yet the problem remains that “AI” is more a genre, or a culmination of centuries of science, than a trend that can be showcased in a single exhibition for public consumption. The result here made it feel like they were biting off more than they could chew.
Eleanor: “I found bits of it very interesting … but … it was a little bit chaotic and at times overwhelming”
We did have a few interesting attempts to chew some of the material ourselves though.
In particular, André and I spent a while discussing how words could help a viewer move from the “cultural” section, which showed our preoccupation with giving life to inanimate objects, to an explanation of the technology itself, which is a separate creation mimicking aspects of “life” that we find so integral to our identity.
Maybe that was the point. Here we were desperately trying to unravel the “mystery of the AI exhibition” by talking about what the story should be. Perhaps when there are so many parts it’s a natural human instinct to piece them together and create our own narrative. Isn’t making connections the strength of our intelligence too?
I started to think more broadly about “more than Human” actually means in this context? Could it also mean “greater than”, “exceeding”, or “in addition to”? The ambiguity was fascinating.
We are collectively failing to find words for this new phenomenon because our language naturally humanises something that just isn’t human. Another synonym is “taller than”... really? One particular sentence that stood out to me explained how we are now making computers that “see, hear and move”. These are human words for human actions. If we don’t understand how entirely differently a computer manages these tasks, the myth and mystery will just continue. André and I tried to reword some of the signs but couldn’t - we all have a way to go.
Walter: “I liked it … but it felt a bit like AI for AI's sake sometimes”
On the whole, this wasn’t what we expected (or hoped for). It was, however, an excellent topic of discussion and a very bonding group activity. In the end we are in England, where complaining is a national sport. Do go and see for yourself, bring your friends, and piece together your own story.
You may find it reassuring to know that at the exhibition, some of the technology didn’t always work and there was a leak in the roof. We aren’t there yet.