Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AI is slowing down

wheresyoured.at

Post reply on HN
Print view View original

771–780 of 820 posts

Re: AI is slowing down

#771

Post by dcre »

afro88 wrote: post #607
avarun wrote: post #578

Earlier quoted context omitted.

There is no substance to the article to tackle. Ed Zitron is a clown that makes seven million dollars a year writing unsubstantiated bullshit on the internet. He has been consistently wrong in every single one of his predictions and will continue predicting the sky is falling, because his content has zero educational value. It is purely a confirmation-bias inducing dopamine hit for the perpetually aggrieved.

I'd love to read about the predictions that have been wrong (genuinely)

He has said every month for the past three years that there is no technical progress left to be made in LLMs and that there is no more room in the market for inference spending to grow.

Here[0] is a fun selection of excerpts from his July 2024 post "How Does OpenAI Survive?"[1]

"I see no signs that the transformer-based architecture can do significantly more than it currently does."

[0]: https://xcancel.com/pathsnotchosen/status/206360940100129633...

[1]: https://www.wheresyoured.at/to-serve-altman/

Re: AI is slowing down

#772

Post by simianwords »

abroszka33 wrote: post #650
simianwords wrote: post #38

Earlier quoted context omitted.

I'm collecting many kinds of predictions Ed Zitron made so that you can see for yourself whether he has a good track record. ------- > While complex, generative AI is a technology that probabilistically generates answers, and has no "intelligence." It is inherently limited by its architecture, and in turn can only get "better" in a linear fashion. I see no signs that the transformer-based architecture can do signific…

> He wrote this in 2024 before reasoning models came out. Remember how ChatGPT was in 2024? Do you think this person is someone who gets predictions right? Any benchmark on AI shows that it gets better linearly. So his prediction was correct. There is no exponential AI improvement anywhere.

>Any benchmark on AI shows that it gets better linearly. So his prediction was correct. There is no exponential AI improvement anywhere.

?? https://metr.org/blog/2025-03-19-measuring-ai-ability-to-com...

Re: AI is slowing down

#773

Post by vatsachak »

LogicFailsMe wrote: post #716
vatsachak wrote: post #381

Earlier quoted context omitted.

Alright, let me explain what's happening this Q Chinese providers realized that LLMs have peaked and have started trying to reduce the price per token. Deepseek pro v4 can easily add tests to my complicated code and costs cents for a million tokens. I can ask Claude or ChatGPT architecture questions and then use Deepseek for the rest. How are these businesses going to pay to price of energy and GPU depreciation again…

I love nonsense like this. If using a larger model to plan a chain of thought task for a smaller model works, what makes you think Anthropic can't do the same thing and offer it as one of their effort level settings? It reminds me of all those fallen AI ASIC startups insisting they can crush Nvidia until they found out the hard way that the rules of the game are dynamic. The real challenge IMO is whether enterprise w…

What stops Anthropic of American energy and infrastructure costs. And those AI ASIC companies just got bought by OpenAI and you guessed it... NVIDIA

Re: AI is slowing down

#774

Post by emil-lp »

simianwords wrote: post #770
emil-lp wrote: post #761

Earlier quoted context omitted.

I reran the numbers, break even after four years. 10k users. Minimum upfront: about $15M Comfortable upfront: $20M–$25M Monthly revenue needed: $900k–$1.5M Required price per 10k individual customers: $99–$149/month API-equivalent output price: usually $8–$20/M output tokens, unless utilization is very high.

dawg -- what is your point? please write your statement clearly and we can discuss.

My apologies for not being clear about my point.

I try to do some napkin math of what it takes to start a company serving an LLM to customers. I thought maybe 10,000 users sounded like a reasonable number.

I'd like also to compare it to traditional Internet companies, e.g. Twitter. I'd guess Twitter with 10k users would cost me literally a dumpster dive, so essentially no cost.

If I were to start an LLM company, what would my initial investment look like?

Turns out around 15+ million dollars assuming I would get 10,000 users willing to pay 200$/month.

I just don't see how the numbers you show add up.

Re: AI is slowing down

#775

Post by padjo »

asveikau wrote: post #120
zachthewf wrote: post #15

Before you spend 20 minutes reading this article, it's worth understanding that the writer has been posting popular but consistently wrong takes for 2+ years (e.g. https://www.wheresyoured.at/peakai/ from March 2024) arguing that AI is failing, is a waste of money, is bad, will never work, etc.

Not sure where I heard this, but I'm reminded of a story about someone predicting the dotcom crash early, circa 1998. For 2 years they were demonstrably crazy, and missed out on massive stock market gains. Then they were right. (And yes, tech slowly bounced back after that.) Predicting the timing of such a thing is notoriously difficult. I don't think being wrong about timing 2 years ago means there won't be a correc…

If I just continually predict that it will rain tomorrow I'll be right eventually. That doesn't mean that I know anything about the weather.

Re: AI is slowing down

#777

Post by zuzululu »

Aerolfos wrote: post #599
latexr wrote: post #583

Earlier quoted context omitted.

> Which is even sillier when you think about it, because if it's useless, then you really shouldn't care: the markets will eventually find out that it's useless, and everything will go back to normal People who are against AI don’t care if it’s useless, they care it’s harmful . And you can’t systematically cause harm then say “oops, our bad” and have everything return to how it was with a snap of the fingers. The con…

> > I do think they're worried about what it'll do to their prestige > Why must this always be the argument? It was the same with cryptocurrencies and NFTs, there is a specific type of proponent who always accuses critics of secretly being pro the technology but publicly against it due to some ulterior motive. Most people aren’t selfish lying rat bastards who think like that. Meanwhile, the prestige to be gained/lost…

been involved with several crypto projects as a dev

i think you are wrong

Re: AI is slowing down

#778

Post by abroszka33 »

simianwords wrote: post #772
abroszka33 wrote: post #650

Earlier quoted context omitted.

> He wrote this in 2024 before reasoning models came out. Remember how ChatGPT was in 2024? Do you think this person is someone who gets predictions right? Any benchmark on AI shows that it gets better linearly. So his prediction was correct. There is no exponential AI improvement anywhere.

>Any benchmark on AI shows that it gets better linearly. So his prediction was correct. There is no exponential AI improvement anywhere. ?? https://metr.org/blog/2025-03-19-measuring-ai-ability-to-com...

Ok, I guess there are some weird benchmarks where they managed to squeeze out an exponential. But even with Google's, OpenAI's or Anthropic's own benchmark it's a linear improvement between models.

Re: AI is slowing down

#779

Post by datsci_est_2015 »

bawolff wrote: post #746
datsci_est_2015 wrote: post #678

Earlier quoted context omitted.

What’s the line that separates grift from speculative investment? Does it only require true believers? If Elizabeth Holmes truly believed in her product, would that have made it not a grift? > The AI situation seems more like a bunch of investors knowingly investing in something very speculative. If they lose their money, while that is the nature of speculative investments. This is a bit reductive, especially as the…

> What’s the line that separates grift from speculative investment? Are they actively lying to people in concrete ways? Not just, we think this is going to be big, when that is questionable, but do they say AI can do X right now, when objectively it cannot even remotely do X under any circumstance? Lying is what makes the difference. > well, maybe there isn’t a word that elegantly describes that situation, but “grift…

To your first point about lying, surely that’s not the bar you’d like to set? Misleading investors is pretty much the greatest talent of all of history’s greatest capitalists - were they all grifters? Is it the extent of the lying that matters? Because pure honesty is nowhere to be found in SV.

> I think the word you are looking for is capitalism.

“It” may be a consequence of some countries’ and cultures’ attempt at implementing capitalist policies, but I don’t think that’s anyone’s working definition of capitalism.

“It” being a runaway concentration of wealth being held by few individuals who are not as well-informed and smart as they think they are, exploiting economies of scale and highly financialized systems.

> That is how the system is supposed to work.

The system is supposed to work in the way that we design it to work. Massive speculative bubbles where every rich yahoo is simply looking for an “exit strategy” is far from ideal.

You’re far too forgiving, covering your eyes while the industrialists rob us blind. There’s a culture of grift and exploitation running rampant in the US at the moment, and this sort of apologia only serves to exacerbate.

Re: AI is slowing down

#780

Post by oudlys »

techblueberry wrote: post #327
asdfasgasdgasdg wrote: post #53

Earlier quoted context omitted.

I can see an argument that the productivity gains are illusory / don’t translate to economic productivity. I’m not denying the possibility. However, most of the engineers I respect have gone from being skeptics a year ago to convinced today. I don’t personally know any true holdouts any more. If there are studies that disprove productivity gains more than six months ago, I’m happy to believe that it was true of the A…

I’m going back to being a holdout, but it’s nuanced - My theory into why LLMs don’t lead to the colloquial definition of productivity would be something like - if code was never the bottleneck than generating code faster doesn’t result in more meaningful output. Even if you take for granted that AI is as good as the best people say in writing code. And Ive spent a lot of time generating codes, I won’t disagree - Then…

>So from a holistic perspective, I think intentionally limiting your own AI usage is the best approach for maximum long-term productivity.

I think this is right. They are much better applied as editors than authors, IMO.

The key thing is stay in control of your output. i.e. understand it thoroguhly. I think you let the LLM make decisions you don't really understand, you're increasing the likelihood of introducing defects that are expensive to address.

Post reply on HN