Which, to be honest, is probably worse. When it comes to AI, I'm a lot more worried about "information pollution" than anything else.

Weird to be worried about a metaphor instead of something real?

What makes it not real?

Imagining that information exists in some pure unvarnished state (like a pristine mountain lake), until it is polluted by some bad actor (big bad logging company).

The metaphor doesn't work at all. It just more "the truth is under attack"

Exactly what I thought when reading this. And that's the moment I stop taking the (probably somewhat valid) claims seriously. This reminds me of the arguments against streaming we had a few years ago, that confused compression/decompression, completely disregarded local caching and calculated as if all network infrastructure in the world was just being used for streaming.

Awesome. We don't need server AI of any kind then.

Always good to have user feedback.

You heard him folks, lets turn off the big servers, local low power models are good enough.

It’s not going to be a good email.

the local model needs to be trained beforehand though, right? And that's where a lot of the energy and water consumption comes from.

They cite consumption during training as a separate bullet point.