> Importantly it was admitted that internal Anthropic models had been used in their proof of Euler blowup; I therefore felt I could not consider Levent to be an independent academic.
If an Anthropic employee is doing independent research, but with models that aren't available to the public (because they're internal models), then . . . idk. It's not clear to me why that should necessarily require a refusal to cooperate between OpenAI and Anthropic employees who are excited about solving a problem like this.
For me, the bigger question here is what "internal models" means to these employees, especially in the context of the OpenAI employees repeatedly avoiding directly answering whether their model had been trained on Tristan's and Levent's ongoing work on the problem. It had always seemed like a loophole that AI companies might be tempted to exploit: yeah, they can say that they won't train on your data, but if an AI company doesn't care about ethics, they might go ahead and train a model for internal use only on everyone's data anyway, just to have as much data as possible and potentially gain an advantage in what the company can internally do. They could never publicly release any versions of a model like that, of course. And of course this is speculation.
This is being reported as OpenAI wanting to strip an Anthropic employee of academic credit for the work they did. What the OpenAI person involved is claiming is that they wanted the outside researcher(s) to put their name on OpenAI's work: to headline OpenAI's publication of what they earnestly believed to be an independent result.
If true, that's generous and beyond the level of generosity one should expect. Extending that courtesy (beyond academic norms) to a competitor is expecting too much. It take a result OpenAI spent millions of dollars on, and put "Anthropic Researcher" right on the cover.
This is, of course, taking OpenAI's side of the story at face value. But it is a consistent, coherent, and ethically justifiable series of events, if indeed it happened that way.
> What the OpenAI person involved is claiming is that they wanted the outside researcher(s) to put their name on OpenAI's work
> If true, that's generous and beyond the level of generosity one should expect
"We highly likely stole your work, and threatened you with 'this is bad for your career' and we refuse to acknowledge any work by your collaborator just because he works at a competitor, but we are so so so so generous"
The two proofs are structurally very different, and don’t even prove the same conjecture. It’s becoming very clear that OpenAI did not steal anything here.
If it was "very clear", OpenAI wouldn't be threatening the researcher with "it's very bad for your career" or state "we can't tell you if it was trained on user input".
Additionally, according to the researcher, OpenAI's proof follows the same approach they used, and which was largely unused in academia, but OpenAI claimes they arrived at it immediately.
This reminds me of a gem of a comment from about a month back, about a dead simple Russian guidance system from a Cold War-era missile: https://news.ycombinator.com/item?id=47389285
Simple spin-scan but with a rolling airframe. The technique is still used today. It is simple only if one looks at individual components. The total package is a different beast.
Authoritarianism has a lot of obvious appeal for the authority, but what is in it for the bootlickers? The answer isn't clear, but I think it mostly rests on a sense that it feels safest/strongest to be a member of the tribe whose alpha is the biggest baddest gorilla. People that support an authoritarian leader often believe that the leader will protect them and be on their side.
Of course, anyone who knows even a bit of history of authoritarianism knows that Dear Leader will happily kick people out at the slightest provocation and seems to always need his subjects to make sacrifices he himself never seems to have to make. But there is always no shortage of suckers who just want to feel like they're on a winning team.
It looks like it's supposed to be a general purpose robot servant, although "Follows you around" from the list of skills on the pre-order page is hilariously disturbing. There would be a market for people willing to pay for a very humanlike robot, for the novelty if nothing else.
Plus if you want to make a robot which can do everything a human can do, without exception, you'll start running into weird edge cases if you make it with a fixed wheeled base. Like suppose you want it to climb a ladder or turn sideways to slip into a narrow space and retrieve something. Or maybe you want it to be able to climb in and out of vehicles designed for human occupants. Would you want to have to buy a special vehicle to bring your robot somewhere with you?
And judging by the name of the company and that they likely aspire towards more than what they've made so far, I'd guess that a long-term goal could be to also make something for people who realize the weakness of their flesh and crave the strength and certainty of steel.
The space race was to the death. Both sides thought that losing meant they'd get nuked or taken over by the other. A narrative like that for Mars is harder. "We'll take over the red planet and come back and take over the earth! . . . Someday!"
For better or for worse, the equivalent race right now is in AI.
> The space race was to the death. Both sides thought that losing meant they'd get nuked or taken over by the other.
... Eh? No they didn't, certainly not by the time people were looking at the moon. By then the nuclear doctrine was very well established. The space race was primarily about _prestige_; it was a propaganda thing. That's why it was dropped so completely once the arbitrary goal was reached; little technology from the Soviet programme and virtually none from the US one were used after both countries reoriented around lower-cost space programmes.
In hindsight, the moon race was pretty close to pure propaganda.
I'm not entirely sure they knew it at the time, though. It was likely a bit of a land grab as well?
The moon is potentially a treasure trove. Bringing anything back to Earth is obviously the challenge because of the energy expenditure but in the 50s and 60s they were considering nuclear-powered craft (Project Orion) and at the time, mining the moon probably seemed like it might have been feasible in the next 50 years.
little technology from the Soviet programme and virtually
none from the US one were used after both countries
reoriented around lower-cost space programmes.
Is this really true? We stopped putting people on the moon, but we certainly kept putting things into space (military and otherwise) and was much of that know-how not directly informed by the technology developed for Apollo? Building life support systems, the rockets themselves, etc.
I'm open to correction/clarification; I'm not an expert obviously.
> but we certainly kept putting things into space (military and otherwise) and was much of that know-how not directly informed by the technology developed for Apollo? Building life support systems, the rockets themselves, etc.
It was _informed_ by it, but, for instance the Saturn V's kerosene and hydrogen engines, developed at absolutely horrendous, unthinkable expense, on the justification that it was the moon race and that the cost would be amortised over expected hundreds or thousands of Saturn launches in the future, were thrown away (reusing the hydrogen one occasionally comes up as an idea, but has gone nowhere so far).
Same for the Soviet stuff, to a large extent; the N-1's engines, also very sophisticated, were essentially abandoned (though a derivative was used a long, long time later in a Soyuz launcher). The Soviets did at least keep the Soyuz orbital vehicle (originally developed for their moon programme), but little else.
This all made sense at the time; there suddenly wasn't much money, so reverting to the less capable, less complex, arguably by then previous-gen launchers was rational. But it was really a demonstration that, by then, neither state took any of this at all seriously (even before Apollo 11, the Soviets had all but abandoned their programme); if it had really been seen as a matter of national security, the Saturn C-5N and N1F and all the rest of the planned evolutions would have launched in their hundreds.
There's a lot of psychological factors in markets swinging up and down, and there are always people saying that there's going to be a recession. Also, if you think about HN specifically: if the market is doing well, great. If the market is doing poorly, then cash may be harder to come by, but there's way more upside for growth as the market recovers. Jumping on a fearmongering bandwagon doesn't fit with the entrepreneurial culture of the site.
Are there any companies that are hybrids of this, where they have some fixed structure in terms of voting power that's, say, 60% employee-owned, and 40% which is publicly traded and has a board? So the employee-owned side of things maintains the primacy of their skin in the game because the 60% control that they have is not something which can be amended, but they can also receive outside investment to some degree? As someone who's definitely not an expert on these things, the idea seems intriguing to me, but I have no idea if there are some reasons why this obviously cannot logically or practically work (as opposed to simply being something which people don't presently do--or something which I just haven't heard of people doing).
You can definitely make these kinds of things work and eg worker cooperatives are generally legal to set up around the world, but they don't necessarily work any better than vanilla companies (and that includes not necessarily being better for the employees).
Worked for one with this setup. It was all great until the stock price fell to ~20 % of its original value and has stayed that way since. Employees didn't like it, which was kinda funny (it's a risk they signed up for).
It's called a population pyramid because the shape of that graph for a growing population is a pyramid. If you look at some of the African countries on that site (like, say, Uganda and the United Republic of Tanzania, if you're starting from the United States page), you can see they're closer to a pyramid shape because they have higher birth rates and higher mortality over time.
Also, on the graph you linked to, you can click on the line in the graph of population versus year to the right, and if you start back in the 1950s and work your way along to the present, you can see the big lump of the baby boomers getting older. That's why people say the United States has an aging population.
> Importantly it was admitted that internal Anthropic models had been used in their proof of Euler blowup; I therefore felt I could not consider Levent to be an independent academic.
If an Anthropic employee is doing independent research, but with models that aren't available to the public (because they're internal models), then . . . idk. It's not clear to me why that should necessarily require a refusal to cooperate between OpenAI and Anthropic employees who are excited about solving a problem like this.
For me, the bigger question here is what "internal models" means to these employees, especially in the context of the OpenAI employees repeatedly avoiding directly answering whether their model had been trained on Tristan's and Levent's ongoing work on the problem. It had always seemed like a loophole that AI companies might be tempted to exploit: yeah, they can say that they won't train on your data, but if an AI company doesn't care about ethics, they might go ahead and train a model for internal use only on everyone's data anyway, just to have as much data as possible and potentially gain an advantage in what the company can internally do. They could never publicly release any versions of a model like that, of course. And of course this is speculation.
reply