Go AI + Humans

I got a little leisure time last Sunday and headed off to my favourite proper coffee joint in the next village with a slightly outdated copy of Wired. What a treat I was in for! The best damn article I have read in ages! It not only describes an important and super exciting milestone in the quest for true AI it’s also wonderfully written- as captivating as a fictional short story (hurrah for real writers!).

The article was The Rise of AI – What the AI behind AlphaGo Can Teach Us About Being Human by Cade Metz in June 2016 edition of Wired. It describes the events that took place in Seoul earlier this year in which Google’s Deep Mind AI “AlphaGo” beat the human Go world champion Lee Sedol in four games out of five. Go is an ancient, complex game of strategy with literally gazillions of different possible game paths. This mean you can’t use brute force computing to mathematically determine every possible outcome like you can in Chess (even a mobile phone can beat a chess champion these days).

It’s one of the reasons why the Machine v Human Go game was truly significant – but it goes deeper than just winning . It’s how AlphaGo won the first game – deliciously described as Move 37 – a move that was stunningly original (the probability of a human making the same move in 1 in 10,000). So the machine not only learned the game but also somehow created it’s own non-human tactic to win. Yeah wow! The how is pretty cool – the “machine” is really an artificial neural network that “learns” using methods inspired by biological central nervous systems. Google has already been using this technology for a while in, among other things, speech and image recognition.

What happens next though is the best bit of the story for me and I don’t want to give away a great climax in the story.

So please read this article – be moved!

What the AI Behind AlphaGo Can Teach Us About Being Human

The article concludes with something I’d never properly considered – the idea of that humans might evolve alongside Artificial Intelligence – and I don’t mean plugging ourselves physically into the network (although that is going to be cool) but the idea that Artificial Intelligence will inspire us to think in new ways and hopefully help us tackle some of huge, complex problems that we face in the world today.

Other references:

https://googleblog.blogspot.ie/2016/03/what-we-learned-in-seoul-with-alphago.html
https://en.wikipedia.org/wiki/Artificial_neural_network
Image: Korean Couple play Go https://en.wikipedia.org/wiki/Go_(game)#/media/File:Korean_Game_from_the_Carpenter_Collection,_ca._1910-1920.jpg