While browsing YouTube videos late one evening a few months ago, I happened upon a presentation about the Google “Alpha Zero” project. If you’ve not heard of it, it’s a Google artificial intelligence project investigating machine learning techniques, and used chess as an example. Cutting a long story short, the system learned to play chess from scratch in four hours to a standard higher than any computer chess program in existence. Various grandmasters analysing it’s games were kind of terrified by what they saw – it appeared to understand positional ideas that took us decades to discover, and essentially played like a human.
I got sucked into watching a series of chess videos over the coming days and weeks, and ended up tripping over a quite wonderful website called “lichess” – allowing people to play each other on the internet for free, with no advertising. It’s years since I last played chess properly, so I started out hilariously badly – and I also recognised that I only typically get the chance to play late at night – when I’m tired, and making mistakes.
Here’s the thing though – I’ve kept at it, and have discovered that I really don’t mind losing at all – it’s become more about playing the game. When I was younger I hated losing at games – but now I appear to just enjoy playing – win or lose. Maybe this is a change children have brought about in me – I’m not really sure.
I suppose the reassuring thing about playing real people is that they make mistakes – when you play a computer you can of course lower it’s playing strength, but typically that resorts in entirely fabricated mistakes, which aren’t quite the same as a human either blundering, or just getting into a complete and utter mess.
I’ll have you know I’m rather good at getting into complete and utter messes. I’m less good at getting back out of them.