Credit: Getty Images
![](https://unherd.com/wp-content/uploads/2019/09/chivers-100x100.jpg)
“If one could devise a successful chess machine, one would seem to have penetrated to the core of human intellectual endeavour.” So claimed an influential 1958 paper about the future of AI.
Then in 1997, Chess Grandmaster Garry Kasparov lost to the IBM’s Deep Blue. Needless to say, the core of human intellectual endeavour remained unpenetrated. Now there are any number of grandmaster-level phone apps around. It seems silly, looking back – the idea that human intellect could be encapsulated in something as constrained and limited as chess. We forget just how huge a task it was for early AI developers.
We are no longer so shocked when computers beat us at games: when the algorithm AlphaGo beat the world’s greatest Go player in 2016, we were surprised but hardly bowled over.
Even though Go is a far more complex game than chess, and even though AlphaGo was a much more “intelligent” player than Deep Blue – it largely taught itself, rather than being taught by humans – we now think of games as something that computers are good at.
But we still don’t think that about the ‘softer’ aspects of human thought. Emotional intelligence, verbal skills. We don’t feel threatened by computers doing things that feel computery, like playing games or recognising images of faces – even though they didn’t feel computery once. But computers having conversations, or writing poetry, that feels different.
Inevitably enough, though, it’s on its way. Elon Musk’s nonprofit OpenAI has just announced a new toy: a text-writing AI which, if you give it a few lines to start it off, will generate an amazingly plausible passage in the style you gave it.
Join the discussion
Join like minded readers that support our journalism by becoming a paid subscriber
To join the discussion in the comments, become a paid subscriber.
Join like minded readers that support our journalism, read unlimited articles and enjoy other subscriber-only benefits.
Subscribe