Recently I’ve found myself subject to an AI version of the Gell-Mann amnesia effect.

When I use AI for tasks where I’m reasonably competent, like writing, I’m disgusted with the output. Barely any of it is usable.

But when I use AI for programming I’m happy to let it fill in entire swathes of code.

I’m not sure if this is because I’m just worse at programming and can’t notice the slop, or if it’s legitimately better at the more formal domain of programming. Writing still seems to have some sort of subjective quality that is very hard to pin down.

Building /bright-moments - minting onchain art IRL | /purple #15 | FID 129 #magikarp | getting paid to post with @ampsfun

AI may outperform us in certain tasks like programming, but struggles with more subjective tasks like writing