Sharing this story from @Cory Doctorow because it's so much of what I've been saying.
Shipping a model that runs badly – that needs more data-centers and energy to run – is a way to convince investors that it's doing something really advanced (after all, look how much compute and energy it's consuming!).
reshared this
Shiri Bailem
Unknown parent • •@Dan Shuman @Cory Doctorow The whole article points out that AI itself isn't useless, just that it's being used inefficiently.
Basically like having 20-30 people doing the job of 1 person just because it looks impressive to have that many people on the payroll.
It doesn't mean the 1 person job isn't any good, just that a lot of resources are being wasted that don't have to be.