disgruntledphd2

> We were 10 years from paying off the national debt when Clinton left office. 10 years!

To be fair, if the US government had continued paying off the debt, the only way the (global) books would've balanced would've been for the US populace and businesses to take on waaaaayyyyyy more debt. Not sure if that would've been better.

(For a much better explanation of this, read Trade Wars are Class Wars).

> leading to World Depression II

This is very, very unlikely. The US would probably suffer more, given how much AI related expenditure there is. The EZ mostly wouldn't notice, and China is already basically in a depression that it's desperately trying to export its way out of (which seems unlikely without them doing something about all the underwater property debt).

> You really end up needing a super majority.

Assuming you're talking about the Senate filibuster, then this can be changed or removed by simple majority vote. It's not in the Constitution, just a rule that the Senate adopted.

Like, personally I think that even sans filibuster, the Senate would move much slower than the House, because of election cycles.

> Oh of course, but data sciency ML type people aren't [supposed to be] tasked with of operating and securing the infra, that would make it even worse.

But AI collapses the need for expertise, right? So the DS people can now do security at the level of the best in the world, right?

Some form of this issue is playing out in lots of businesses right now, and I'd be very surprised if it wasn't happening in the labs.

I suspect the lab's infra people are focused on inference, rather than evaluation. I agree that this is insane, but it seems like the most likely explanation to me.

For converting unstructured data into (more) structured data, they're phenomenal.

Yes, they are expensive, but flexibility has a lot of value.

> Even the largest, most trusted companies in the world get regularly hacked

Yes, they do. Fundamentally this is because they don't spend enough money on security (like basically everybody else).

Some form of personal liability for executives is one (relatively simple) way to ensure that this becomes a priority, like SOX did for financial reporting.

There are other ways, but I do like the personal liability approach as its targeted, and has been used successfully in the past.

> I know that the people working there aren't idiots so the most likely explanation is that the incredibly weak security was intentional because its inevitable breach would make for great marketing.

It's more likely to be different specialisations. Most of the people doing the evaluations are more data sciencey ML type people, rather than software engineers. This isn't helped by their culture which is very much driven towards alignment as the only possible solution to super-intelligence (which may be true, but I have my doubts that this will happen in any reasonable time frame).

Yeah, because I was told that it was mostly trilobites, around the same time (mid late 80s).

> At least Europe has better weather and social services.

Europe is not a country, these vary wildly across members of the EU/UK/Europe more broadly.

As an Irish person, I'd love if the weather and social services were better here.

> Manufacturing offshoring started in late 90s

Speaking as an Irish person who grew up in the 80's, this isn't true as there were loads of US companies manufacturing in Ireland back then. There were even clothes manufacturers until the mid 90s.

So I'd probably agree more with the OP here, perhaps the Asian outsourcing happened in the 90s (but even then, loads of my toys as a kid were made in Taiwan), but this has been happening for a while.

It's potentially related to the relaxation of capital controls that happened in the 70s.

Much, much worse for the EU, probably about the same for China (based on their actions so far).

This is the bitter lesson that the labs may have to learn for themselves. Like, even with AGI, without corresponding enhancements in robotics, the machines basically _can't_ take over. Intelligence is not all you need (you also need hands).

You see this in Dario's appearances, where he jumps from AI is doing well using verifiable rewards (true), to therefore we can automate basically all science (maybe, but not without a lot more work in how it interacts with the physical world).

The labs have raised far, far more money than any of the hyperscalers ever did, with no sign of a route to monopoly profits.

Yeah, we should give a limited copyright waiver for pre training, given that the model is released as open weight and allow labs to compete with RL on that basis.

Yeah, it's generally the harness around the model that makes it useful.

Presumably they used intent to treat though, so 73 pp data were analysed.

Hey, I supported it too! I actually thought it was her best decision.

It hasn't worked out politically, though.

Again, I'm aware, I just wish different decisions had been made.

Even Bibi has been screwed by this war, because it hasn't made Israel safer, and he's got an election coming up.

> If there was an actual spark, it was the 2016 election as this was the first time when all the disinformation, professional troll farms and Russian influence truly came into play on a world changing scale.

I'll never forget how insane HN went in 2016, felt like basically the whole English language internet went insane. And it's been a decade since then (wow).

> This was the desired effect though

Honestly, I think you're giving far too much credit to the current US government.

> In a world where the Russo-Ukrainian conflict doesn't happen she was fine.

The German (so not just her) position on the post financial crisis reforms were wildly misguided, and her taking in of Syrian immigrants also didn't end well.

I think that shutting down the nuclear plants would also have been recognised as a bad idea, even without the war.

> And it doesn't help most of our economies are all export economies,

The worst part of all this is that becoming more export focused was pushed heavily by Germany during the euro crisis, and now its causing more problems.

Honestly, I'm coming round to the notion that Angela Merkel was literally wrong on everything she ever touched.

Mind you, German political culture consistently makes the wrong decisions (basically in hock to the ICE car industry, which has aged badly).

Wasn't this during the period where they had a bunch of bugs around caching and the models were making loads of weird decisions?

I honestly feel like basically nobody knows anything about these models, it's all just vibes (and I'm no different).

It's worth noting that Cochrane reviews are normally the minimal effect sizes you'll see, as they pull in every study and hate self reported measures.

Don't get me wrong, they're my first port of call (mostly for the funnel plots), but they're one source, with their own biases.

The reason that algorithmic feeds were introduced was essentially that Zynga and various heavy posters were killing Facebook (but mostly Zynga).

Back near the dawn of time, FB's feed was chronological, and that meant that Every.Single.Farmville.Notification showed up in your feed (and there were a lot of Farmville notifications).

This was actually causing real problems for Facebook, and they introduced an algorithmic feed to fix it. Back then, it was actually relatively simple, and mostly just ranked posts based on how connected you were to the person on Facebook.

Obviously things have changed since then, and algorithmic feeds have had a host of weird impacts, but a ranked feed is not necessarily bad in and of itself.

> (though I'm just starting to try WSL and so far it is looking good)

WSL is really nice, as is Windows Terminal. I particularly love its fonts (but that's probably just me).

> I feel like the evidence standard has to be extremely high if you're claiming that tens of millions of people are going to die in the next few years with the goal of blocking a program from being defunded.

I mean, this is basically the evidence standard that we use for flu/respiratory illness deaths, along with lots of other hard to measure interventions.

Personally, even if there were no deaths due to USAID being removed, it would still have been a strategically bad thing for the US government to do, in terms of soft power.

Additionally, the US are noted in development circles for circular aid, in that they gave a lot, but then required it all to be spent on US goods, so it was more of a subsidy for domestic producers.

I do agree that lots of people just jump on these numbers uncritically, because of prior beliefs about Musk/Trump/DOGE, but it would be very very surprising if there weren't lots of avoidable deaths due to the cutting off of aid. Like, the same thing could have been accomplished in a much better way by tapering off the aid as programmes expired.

You probably shouldn't believe leaks when a company has a financial incentive to appear profitable (particularly when attempting to IPO).

> The point is that the model is already immensely profitable. There is no fundamental reason why AI isn't profitable. It already is.

We'll see I suppose. Until the audited financial statements are released, no-one who doesn't work for an AI lab can be sure.

> Kimi K2 Thinking was apparently build by just 80 people and not more funding.

But they're not competing (offically at least) for GPUs and they have much, much cheaper electricity (which makes pre-training more affordable).