Post

Thank you for sharing. It makes a lot of sense to focus on "how and why some people may use this technology to cause great economic harm" instead of "this technology causes harm." It's not an LLM's fault if for-profit hospital owners want to use it to cut costs instead of improve outcomes.

Weather models are usually just atmospheric models (with land). Climate models need to have ocean and sea ice components as well.

Weather models are usually just atmospheric models (with land). Climate models need to have ocean and sea ice components as well.

NeuralGCM is an atmospheric model, not a climate model (despite the headline). Climate scientists are busy building GPU friendly models using C++, like the DOE SCREAM model. But we still have ways to go.

Really good, thanks for sharing.

True. But how do we know that the human brain just isn't one of these, too? The only arguments I've heard for the brain being qualitatively different all sounded quite similar to "I just know it".

either LLMs are a worthy step on the path to AGI, or they are a local and extraordinarily-wasteful maxima we'll know when AGI is achieved, and probably no sooner :) but I think folks in both camps should "devil's advocate" their own thinking, what if the other folks are right?

There is no recording of the talk?!

Absolutely... But you can't freeze frontier models where they are today and still get sufficient profits to justify existing capex. You have to train new ones sufficiently better than existing ones to justify capex, and once the bubble bursts these hype men won't be able to.

I dared to criticize two professors this week for their facile anti-AI arguments and was told I am a "shill" and a "ChatGPT addict". Reading some of their anti-AI arguments makes me think maybe ChatGPT really would do a better job 🙄 To me, this article is compelling and frightening.

A great read. Major props to UW (Go Huskies) for hosting Cory Doctorow, many amazing thinkers come out of that place. I enjoyed reading enshittification and look forward to reading the forthcoming “Reverse Centaur's Guide to Life After AI”. Thanks for sharing the piece, Simon!

The fact Fiverr existed way before AI is enough evidence that the cost of AI image and video generation probably doesn't outweigh the cost savings of it 🥲 He also has an interesting point about how to pop the bubble that I feel wasn't quite finished.

Employees as “reverse centaurs” is the culmination of Marx’s thoughts on worker alienation. To me, even worse is the “life alienation” we all experience as consumers. Every time you “listen carefully because our options have changed” you make yourself a reverse centaur. We serve corporate machines.

I agree with this almost in its entirety. I would add that some of the criticisms are endemic to our corporatist economy and not just the AI industry. Our western democratic economic system cannot withstand oligarchs or multi-trillion dollar corporations, in the AI field or anywhere else.