The imperious march of technology is no better summed-up than by the advances made in the area of artificial intelligence in recent years. Long mooted and oft-discussed, it is something whose long gestation has garnered pages of analysis and peculiar visions of the future by commentators and artists alike.
From within the chamber of Google’s DeepMind project, a breakthrough was observed by filmmaker Greg Kohs (Song Sung Blue) when Google decided to tackle the ancient Chinese game of ‘Go’: a game involving marbles and a board, and with more move possibilities than there are atoms in the universe. Faced with the fierce intellect and mastery of the world champion, Lee Sedol, artificial intelligence was put to task in a best-of-seven battle to see who, and what, would come out on top. The lessons learnt from the exercise set reverberations beyond techy circles and into the world at large.
We spoke to filmmaker Greg Kohs about his time on hand for this historic event.
How did you come by the project?
I knew nothing about AI at all. Zero. And nothing about ‘Go’ at all. I was invited, or asked, by some folks in Google Creative Lab, which does a lot of audio and video type of work for Google, to help them document this event for their archives. No expectation of a film, but just to document it and maybe they would make something small out of it. Once we started filming, and started to meet the people involved, I just really connected with them a lot. I felt there was a much bigger story, so I proposed to make a much bigger story out of it. DeepMind approved that. As a result, they gave us the resources to be able to make the film.
This film could potentially be preserved as a time capsule piece, as what you’ve captured is a bit of a David v Goliath battle between humankind and computer.
We all thought about the idea that Lee Sedol (Go world champion) has spent almost all of his life – I mean, he started at the age of 4 – he lived and breathed ‘Go’ to become the master that he is and AlphaGo spent 2 years to gain its knowledge. So often people have referred to it as ‘man versus machine’. What I took away from it, and it’s said in the film through the DeepMind team, is that it’s man with machine. I guess that’s where I find it hopeful. For the ‘Go’ community and if you just take that one element, they’re not frightened by it: they’re super excited by it. To learn from it and unlock knowledge and wisdom. I’m told there are AlphaGo inspired moves in the games now.
If you take that and apply it to the real world, if you’re looking to solve problems… cure diseases etc, I would think that in the same way that it’s assisting the game of ‘Go’ and that deeper knowledge, perhaps it can assist a doctor. It won’t give the answer, but maybe it will suggest a pathway to an answer that no one’s thought of. There could literally be more ‘move 37’s’ in the world (the watershed move made in the film by the programme that stuns the ‘Go’ community). I buy into that. That’s what I really felt.
It sounds as though you recognised early on in the process that this was going to be quite important.
This may sound grandiose, but the moment that I felt that it was ‘on’ – that this was important and history (in the making) – was the first move. I tried to communicate what that felt like in the room, because AlphaGo took a long time to make its move. There was this nervousness and everyone was commenting on it. We had to be very quiet. Once that move happened, and it appeared on the screen, it was history. It was like it time-stamped it. It was powerful.
I guess that the biggest hurdle that AI developers face is trying to convince people that artificial intelligence doesn’t mean that computers become self-aware and want to eliminate people.
Yeah, and we try to address that in the film. There are films just about that. This was a different type of film, where we talked about it, because it’s a question that comes up.
Did you pick up on any conflicting emotions from the DeepMind generally about what they were doing?
They were excited, but they also cared about Lee Sedol, because they are human. I didn’t necessarily feel that they were conflicted though.
Three or four days ago, it was formally announced the creation of ‘AI and Ethics’. It has been officially created and announced. That’s the other side. The excitement of creating, but creating responsibly.
Have you watched this film with many audiences?
It’s interesting to take the temperature of the audience and what kind of audience it is. With test screenings, I would always want to make sure that the film keeps leaning forward. I would say to my editor, ‘snip this out, snip that out’.
There’s what I call ‘the ass-lifting moment’ with an audience. It’s normally Act 2, where all of a sudden everyone shifts (in their seat). If I see that, I’m like ‘aghhh’. If I still have control, I make a note to go in and fix it. I do just want my films to lean forward.
As a filmmaker, do you think you might step back and look at this again in five or ten years’ time in order to see where we’re at with it?
I may. The types of stories that I usually tell aren’t topic-driven, they’re people-driven. If there was a human element that was interesting, then of course.
AlphaGo is available through Amazon Prime, iTunes, Google Play now. You can find our review here.