I’ve covered a lot of DC debates over what AI chip policy *should be*
But this week, the argument is over what the policy *has been*
Paywall-free explanation of the Nvidia China loophole drama: https://t.co/dHe2bZGcEP
@capitolhunters@impoliticeye@katz_morris Please investigate Morris lying on job apps about a college degree and/or if campaigns looked the other way, given that 1) he lied about his age, saying he was two years older than he really was, and 2) he never graduated college. Claim of 2 years older -> obfuscate timeline
@CardioNP@kglightspire@RantyAmyCurtis That would explain why he lied about being two years older than he really was. I wouldn't be surprised if he lied about having graduated college
@LeahLibresco Morris Katz lied about his age, which makes me wonder if he also lied about having graduated college (he's a dropout). Seems like the truth isn't a big deal to him
@kyliebytes@tab_delete ... Theo because, um, his parents are two of the biggest fish in journalism? Idk man, wish Theo could be a little more self-reflective and forthright about his ridiculous privilege while making much of a book (and movie licensing rights) about his peers' outsized opportunities
@kyliebytes@tab_delete former student... waits two years until after graduating until a competent reporter shows up, as if the other Stanford Daily reporters weren't also covering lots of enterprising stories. Again, did this student know Theo from tony Andover? And/or how much did the student choose..
@eshear@NathanpmYoung Many EA's don't even care about making the world better. I find them to be the among most competitive people, and they view "make the world better" as a dimension on which they can compete.
@AndrewCurran_ I don’t see anyone discussing the possibility of AI misuse, which would happen before the singularity. If open source models -> widespread misuse -> bioterrorism etc, ppl will want stocked-up bunkers. Only the super rich will be able to afford that. Wealth would enable survival
@nickcammarata perhaps it's because they work in ai and feel like the smart AI models are extensions of themselves that they built and therefore can incorporate into their self worth
other hand, for China right now, there are just some things money can't buy, and that's Nvidia chips. A chance to change that by paying loads of money (i.e., to Iran) seems like a no brainer. In other words, for Trump: Iran > AI; China: AI > money; Iran: money + China relations
Would be shocked if Trump has AI models meaningfully vetted. Feel like so much of the trump admin ai talk has been ppl like @deanwball writing working papers or plans or whatnot but then Trump doing something totally different. And for trump to do a sudden 180 on the issue…
New York Times:
Trump administration is considering the introduction of "government oversight" over new AI models, which could include "a formal government review process for AI models". The process could be similar to the one being developed in the UK "which has assigned several government bodies to ensure that AI models meet certain safety standards".
A working group of government officials and industry executives would be brought together to discuss potential "oversight" procedures, including which U.S. government agencies would provide the "oversight" and whether there is a role in this process for CAISI.
"The White House wants to avoid any political repercussions if a devastating AI-enabled cyberattack were to occur, people in the tech industry and the administration said."
Prediction 2: Trump sells out the U.S. by sending China more Nvidia chips in exchange for China arranging a peace deal (e.g., China paying off and/or pressuring Iran). Trump has shown he hardly cares AI policy, and he would do anything for a silver bullet solution to Iran. On the