Finally writing up my thoughts on ChatGPT. This wasn’t written by ChatGPT (yet).
Oh God it’s boring isn’t it? Who’d have thought that passing the Turing test could be so dull? What was I expecting? Fireworks! Parades! Excitement! 🥳🥳🥳🥳 Instead we got automating the reply guy*. I guess we get the tech-kicks we deserve in a not-very-post-pandemic world: mild disappointment and tired resignation to the corrosion of society. (I cheer up in a moment I promise.)
I’m reminded of something I read by Kevin Kelly too long ago to be able to google the reference, he was talking about our relationship to advancing technology. There are always some test – some horizon – that marks us out as different from computers. They will never be able to do [this thing that’s the essense of humanity]. Once upon a time that thing was playing chess. Guess what? We all lived happily ever after. His point was that we discover that that thing that defines humanity is a mirage soon as computers can do that thing. Phew! What’s defines us now? Oh crap, computers have just done that too. Art, writing, music. The Turing test. Just a mirage. (I’m too scared to mention love.)
The other thing I took from that Kevin Kelly piece was the idea of centaurs: half human, half computer. I’m not sure if its still true but at the time the most powerful chess players were teams of a human chess Grand Master and a chess computer. This is a powerful idea, it frames the challenge not as what the tech can do by itself, but what how the tech can give us super powers – how it extends our abilities.
Indulge me a moment. [Scratch rewind sound] Rewind a few hundred thousand years to a bit before humans were starting to evolve. Rewind to the start of design and technology, to when our primitive ancestors were cracking open flint stones to extend their ability from hit things and cut flesh better than ripping with their fingers. Then they attached their rocks to sticks to extend their abilities beyond their limbs with spears. We can draw a line through time, society, evolution from pre-humans extending their abilities with crude tools to us using ChatGPT. Design and technology is here to extend our human abilities.
How does ChatGPT extend our abilities? (Or does it make us dumber?) At the moment it’s a very uninteresting chatbot. It’s uninteresting because the tool that OpenAI have created is simply about showing off the tech, and the tech by itself is only capable of automating banalities with a high risk of those banalities being factually wrong. But what could it be? Can it be turned into a centaur-like writing co-pilot tool? Almost certainly yes. That’s a bit more exciting. How will we use it give us super-powers? There’s a nice clue here:
This is lovely and warms my cold, worn out tech-heart 🥰 Let’s have less automated reply guy and more of this wonderful, enabling, tech.
A final thought; how will ChatGPT and friends change language itself. The massively powerful chess computers have changed how humans play, and singers have started to sing like autotune. Are we going to write and talk like ChatGPT? Probably.
*I meant to link to an article here but I can’t find it, if I do I’ll add the link. The basic premise was the ChatGPT pronounces with overconfident authority on subjects that it has little knowledge and is often just plain wrong. We all know people like that.