561–570 of 718 posts
How does gobbledygook like this get traction on HN? What has happened on HN culturally to allow something like this to surface to the top?
Disclaimer: I didn't vote for this submission.
The winner in the long term will be the one that will deliver the best performance and low-memory ratio for local models. Anthropic, OpenAI and Mistral are just companies that are making money right now (still not profitable), but will lost their tractions and values in the long term. However, I am more appealing to see how OpenCode Go subscriptions will go in the future: cheaper than big techs, more tokens, and they…
Local models will never compete with large SOTA models, in the same way an iPhone doesn't compete with supercomputers doing nuclear simulations. They paths will differentiate and split. Probably SOTA models will eventually be locked down and only accessible to state actors because of how expensive they will be to run (already started with Mythos).
that might be true for us based providers but i dont see china turning closed source anytime soon.
a lot of chinese labs come from big non ai focused cloud services (alibaba, tencent, huawei) who want new models with higher benchmark scores and lower inference cost. they dont care if the competition gets better because its all open so they can build off each others tech, and if anything happens they got other profitable services to fall back on instead of depend on llms only like anthropic.
also the business culture is way different, in vc backed america you would get laughed out the room for saying "there is no moat we just do the same thing as everyone but better". you need to show infinite potential growth and lock everything down to prevent competition but you can get millions to start with no customers and no profits. in china its all about the real money they dont care if your margin is 10 or 90 percent as long as you stay profitable. the llm providers are profitable so they keep their business model.
That doesn't count as winning at all.
Earlier quoted context omitted.
> You buy an iPhone even though the cheap-o Wal-Mart Android phone for $100 "does the same thing". People buy iPhones because of status signalling and network effects, neither of which appears to apply to AI model choice. LLMs are already rapidly on the way to being interchangeable commodities.
No they don't, it's not 2008. Anybody off the street can get an iPhone or a free iPhone with a mobile plan. They're commodity products. Even homeless people have them. To the extent LLMs are commodity products you're right (so far), but that is limited to the main model providers, such as ChatGPT, Claude, Gemini, &c. with interoperability on cloud platform providers and other technology providers like an Apple offeri…
https://mashable.com/article/apple-messages-green-doj
https://www.sfgate.com/tech/article/apple-green-bubble-messa...
Earlier quoted context omitted.
This is a war because the media says it's a war. The media says it's a war because AI companies are paying them to say it's a war [0]. When AGI comes the threat won't be from which primate turned it on, but from how well AGI is aligned with humanity. All of the war talk is to distract from the alignment problem and instead force investment in hardware infrastructure. [0] https://www.wired.com/story/super-pac-backed-b…
>The media says it's a war because AI companies are paying them to say it's a war [0] When AGI comes the threat won't be from which primate turned it on, but from how well AGI is aligned with humanity. And when the AGI comes, they won't unleash it to defeat US enemies, they'll first unleash it to make more US workers redundant and boost their stock valuation.
Earlier quoted context omitted.
> the main question for me is, why is this a war? It's a war because the hinted promise behind the hype that the first organization to reach some as-yet-entirely-theoretical AGI that can bootstrap itself to godlike capabilities will then Install Planetary Overlord* and rule the world as near-deities themselves, with the rest of the (surviving) human race as their slaves. I think it's a nonsensical idea, but that's th…
>Planetary Overlord* AGI is nice, yet not necessary. The orbit filled with Starlink descendants and datacenters will be the it. Anybody else wanting to get there would have to get permission. SpaceX/Musk have all the components for it to happen - from Starship to AI (including the army of robots on the ground). The governmental power/sovereignty of US will be used as a stepping stone (that is the strategy described i…
The USA, China, and Russia have all successfully tested anti-satellite weapons. If anything, any company that operates a constellation of space-based data centres would need 'permission' to keep them working.
Thats like Microsoft saying "Don't use Linux because selling an operating system is what matters"
Earlier quoted context omitted.
Researches at top AI labs don't consider EY to be a kook even though they may not necessarily agree. EY concepts/terminology appear in Anthropic safety papers. Geoffrey Hinton takes him quite seriously and mentions him in his interviews.
Anthropic is the AI doomer / safetyism lab, and Hinton is one of the patron saints of 'rationalist' AI doomerism. AI doomerism is psychologically attractive to "people with autistic cognitive traits, including dichotomous (black-and-white) thinking, intolerance of uncertainty, and a tendency toward catastrophizing". They are pascal's mugging themselves, to ironically use one of their terms. It's fundamentally a cogni…
"What if AI doom is all fear-mongering, and we create AI less prone to make up dangerous stuff or mistake buggy goals for real ones" (which is what alignment is) "for nothing?"
Even if Yudkowsky is autistic, you're muddling the condition. Autistic people have a *practical* intolerance of uncertainty in the moment (everything unexpected from a noise to a missed turn can be a jump-scare in their day-to-day activities), but often they're absolutely fine with intellectual uncertainty, unconventional ideas, abstract ambiguity, nonconformity, etc. Indeed, one of Yudkowsky's main things is Bayesianism, i.e. being precise about uncertainty.
Yudkowsky's reported P(doom) is somewhere around 90%, which is very much in the realm of "we might eventually be able to figure this out, *but we're not even close to ready so for the love of everything slow down so we can figure this all out*"; the book title comes from a long tradition of authors noticing you need to beat readers over the head with your point for them to notice it.
Anthropic (like at least also OpenAI), appears to think they can solve the problems that Yudkowsky has found. They're a lot more optimistic than him, but they take these problems seriously.
For his work on AI, Hinton got a Nobel prize in Physics, a Turing Award, the inaugural Rumelhart Prize, a Princess of Asturias Award, a VinFuture Prize, and a Queen Elizabeth Prize for Engineering. Calling him a "patron saint" of "doomerism" is like calling Paul Krugman (Nobel laureate in Economics) a patron saint of "Trump Derangement Syndrome" on the basis of what he says in his YouTube channel: a smart person's considered opinions are worth listening to even if you have not got time for the details, because you can be sure someone else has considered the details and will absolutely be responding to even an i missing a dot.
A Pascal's mugging would be more like S-risk (S stands for suffering) than doom risk: https://en.wikipedia.org/wiki/Risk_of_astronomical_suffering
AI is not some divine creation. It is built by humans. History has repeatedly shown that China is able to catch up, and often surpass others in the end.
The article never really explains why AI is supposedly so unique that it guarantees the United States will inevitably win.
Opening up comments to see top comments are 90% "NO U" without any substantial discussion - you disappoint me, HN.