Thread by @Wolven

Listening to Josh Hawley fear-hype GPTs and LLMs at 10am on a Tuesday is not a fate I'd wish on anyone, but this is the life and careers I've chosen.

https://twitter.com/cspan/status/1658474468489203715

I'll say this for Sam Altman: He's definitely learned how to package "I'm deeply concerned about AI overlords and being hunted in a Terminator-esque wasteland of bombed out cites and mountains of human skulls" into a mainstream-appealed senate-hearing soundbyte.
#SenateAIHearing

Thank you, Senator Welch, for helping "AI" and silicon valley sell their model and vision of regulatory capture.

You need to pay VERY CAREFUL ATTENTION to the fact that every plea for regulation is followed by a "…and we would love to partner with you on that."
#SenateAIHearing

Marsha Blackburn coming at "AI" Art mainly from the side of copyright and only mentioning as an afterthought the idea of consent by human artists to have their works used as training data is one of the least surprising lines of questioning thus far today.
#SenateAIHearing

Klobuchar and Hawley both really focusing in on misinformation and democracy; Hawley talking a lot about deepfake-type problems, Klobuchar on generating things like false polling place information, or quickly made rafts of social media posts with disinformation. #SenateAIHearing

Gary Marcus keeps hammering on transparency, but thus far there's only been a couple of brief mentions about meaningful accountability & oversight to change what's gone wrong. Even external boards & regs only get so much if the companies themselves shaped them. #SenateAIHearing

Graham pushing for a regulatory agency and licensure and while also not letting people actually speak and elaborate on points demonstrates how he continues to be the absolute worst. Like, he can very definitely think about complex things, but he is just an bully.
#SenateAIHearing

Coons asking surprisingly trenchant questions about the values of the data on which "AI" is trained, and differences between fine-tuning and constitutional modeling.
#SenateAIHearing

Thread broke:

https://twitter.com/Wolven/status/1658497099237842948