March 10, 2023•289 words
Here's a sentence I read on LinkedIn which made me chuckle:
"AI already outperforms humans on many tasks - like chess..."
I suck at chess. You don't need AI to outperform me at chess. If your fridge has a computer, I'd probably lose to your fridge at chess.
But just this morning, I was excited at the thought of contacting a chess coach and booking some lessons with him.
I had a warm, fuzzy feeling of finding a chess coach from my home town - and realizing he might have played with a friend of mine, back in the day.
And after I've looked through that guy's Lichess record, and saw many draws and losses, I didn't give up on contacting him, just because he was "outperformed". On the contrary: I thought, "wow, this dude must have learned a lot from all the games he lost."
It looks like we're only now beginning to find useful ways to talk about AI.
If these conversations keep treating everything as a "task" at which AI's goal is to "outperform" humans, then I don't want to be having such conversations. And if you're stuck in that frame of mind, I'll let you debate the finer points of it with the chatbot in your fridge until it outperforms you - as I'm sure it will.
I want to talk about the areas in which humans reserve the right to suck, and to love that they suck.
I want to talk about all the things we learn while being outperformed, outclassed, overwhelmed. I want to talk about the rites of passage.
I want to talk about all the possible opposites of the words "task" and "outperform" - and to derive my methodology from there.