@brubarian@WillManidis A lot of the transcripts are similar to public interviews and essays. Most of the alpha comes from the private audience questions imo
The sad thing is I actually agree that a lot of American manufacturing capacity and/or IP was given up far too cheaply for gains that, both in a macro sense, were negligible or negative and even at the individual company level were just not that salient and can be rightly attributed to greed.
I just think that it's a mistake to map on things like pharmaceutical manufacturing or cheap electrical components and what's happening with AI. They're just really different categories and people in DC don't know that. They see it as kind of a way of taking revenge upon bad policies from the 1980s, '90s, and 2000s. That's a nice narrative arc. It's something that people who are policy-wise, who are political and historically literate, are interested in doing.
Technically and substantively they're completely different. It makes no sense and it's kind of ridiculous. I understand the appeal emotionally and as someone who's been worried about the CCP before, it was completely bipartisan to the point of being somewhat scapegoatish but it's just not going to work as currently constructed. The U.S. simply doesn't have the cards, to quote a phrase , in the long term. Big fan tho
Fair . Honestly Chris, I respect you way too much to dismiss you as a neophyte -DC hack trying to make a policy career, spouting about AI. You don't need that .but I really think the supply chain implication of trying to strangle China's AI future is deeply underappreciated by people who share your position. I also dispute the particulars of the Jensen case but that's somewhat beside the larger point. If AI is so big and so powerful, it has to be treated as an almost military-like power( I don't necessarily believe that but this is the internal and implicit justification for these measuress ) Then I think Chinas in a good position to simply blow up the chessboard cuttinh the US off the semiconductor and data center component supply chain. At some point the picture just sort of breaks down. The US simply doesn't have the veto hawks wish it had
@not_ellington@SamyazaSpeaks I think he probably was 27. He served in the war (GI bill) and he had to go to night school, I think, to get all the credits to get to Harvard so I do think he finished a bit later
In fact exit poling shows that among those who cited corruption as an animating issue in the election ,they split almost evenly between parties so there's no evidence that this is something that the trump administration was uniquely vulnerable on. secondly, before you quote tweet dumb Nyt articles about Howard lutnik was CEO of cantor , all these articles and this whole genre of articles tend to demonstrate is that ceos or people involved in high finance when they go into government is that their existing firms continue to be on one side or another of an existing regulatory and or financial / technology issue which is literally inevitable if business people who operate a national international scale are going to participate in U.S Government
@Scholars_Stage There is literally no evidence of corruption here. Zero . And u might want to double check gas prices and Annualised inflation rates before declaring "corruption" the key issue of a federal election.
The thing they don't understand, because they're completely uncurious about technology, is that if the promises of AI are as true as they say, what they're doing is effectively declaring preemptive war on China.
Think about it: at what point does this logic end? Are you going to deny M6 Apple chips to China? Does any computer processing power to China incrementally increase their chances of taking over Taiwan? It becomes a bit of a joke but as we see in the Middle East and also in Ukraine, Western policymakers are not able to think strategically at any level at all.
Disappointing but not surprising
@jonfortt people love to externalise their own flaws. and i know no one cares anymore but intel comm. leaking to the press ought to be criminal and actively prosecuted ffs. deeply unserious people. honestly think its an anthropic op
If you arbitrarily decide to change the composition of the index according to the latest whims of big labs, then by definition open models will always be in catch-up. There's not currently anything like the usage growth in science or electrical engineering for LLMs, like there's been in coding. So it seems somewhat arbitrary to make those the new focus of the index.