A.I’m so excited
This weekend, Kier told me he’d followed my recommendation and started reading one of my favourite Iain M Banks books, Excession. I always love it when people take my advice on books and films. I like sharing the things that make me happy and it gives me a chance to talk about it with them when they’re finished.
But this is a little different. I’m more excited about this than usual because it gives me a chance to nerd-out over something I’ve been obsessed with since I was a kid. You see, most of the characters in Excession aren’t people, or even aliens, but hyper-intelligent machines called “Minds”. Or, in simple terms, Artificial Intelligence.
Now there’s a lot going on with this at the moment back in the real world, and so much of it is straight-up Science Fiction becoming Science Fact. Here’s a random selection…
This is either “My Favourite Film” or “One Of My Favourite Films” depending on when you ask. It’s based on a Phillip K Dick short story, a neo-noir set in a dystopian near-future LA, following a retired police officer called back onto the job to track down the most advanced androids (“replicants”) of their kind. If you aren’t into Sci-Fi, that might sound awful, but it’s more than a story. It’s a discussion on the question:
What makes us human?
This kind of debate isn’t new to philosophical literature, but now it’s entering mainstream conversation. And it’s giving me chills every time it does. For example:
The sequel, Blade Runner 2049, is coming out in October, and I hope it adds to the discussion. Plus, Denis Villeneuve (Arrival, Sicario, Prisoners) is directing, and the teaser trailer looks ridiculously cool, so I’m trying not to get too excited. I’m failing.
Hitchhiker’s Guide to the Galaxy
In this phenomenal series by the late Douglas Adams, there’s a subplot where a civilization takes thousands of years to create the most advanced supercomputer ever, trying to find the answer to the “Ultimate Question of Life, the Universe, and Everything”.
Spoiler alert: The answer is “42”
The reason for this is that they were asking the wrong question. They built a machine without knowing what it did. So, in the story, they take millennia more to build another, bigger and more complex computer to work out what this question was. I won’t spoil it by telling you what that second computer was, because you need to read the book. It’s great.
But in the real world, the people making hyperintelligent supercomputers definitely know what their machines do. You’d think so, right?
Wrong. Watch this to see the Google and NASA Quantum AI lab crew avoiding answering the question of what this thing actually does:
Terminator/The Matrix etc.
Aaaand then there’s “Unsupervised Learning”. This is where starts to get really nuts.
The most simplistic machine learning allows us to teach a machine to look at a cat photo and register “cat” as opposed to, say, “beaver”. Or “discarded wig”.
(Example taken from this article)
Currently, humans need to feed millions of photos tagged as “cat” into the machine until it recognises the features of a cat. But eventually we won’t have to. Eventually, machines will be able to teach themselves, hence “Unsupervised Learning”. That’s the kind of thing they are working on at Google Brain.
Think about that for a second. We are teaching machines to learn not to need us.
If you’re wondering what happens now, this is the bit just before Skynet takes over. Or maybe those weird squid things from The Matrix. But it’s okay, because Arnold or Keanu will be here to save us, right?