My current intuition on this topic is that they are right about scaling but they are training on the wrong data.
LLMs were not intended to be the core foundation of artificial intelligence but an experiment around deep learning and language. Its success was an almost accidental byproduct of the availability of large amount of structured data to train from and the natural human bias to be tricked by language (Eliza effect).
But human language itself is quite weak from a cognitive perspective and we end up with an extremely broad but shallow and brittle model. The recent and extremely costly attempts to build reasoning around don't seem much more promising than using a lot of hardcoded heuristics, basically ignoring the bitter lesson.
I've seen many argue that a real human level AI should be trained from real-world experience, I am not sure this is true, but training should likely start from lower-level data than language, still using tokens and huge scale, and probably deeper networks.
rossdavidh 14 minutes ago [-]
What happens is they go out of business: "these firms spent five hundred and sixty billion dollars on A.I.-related capital expenditures in the past eighteen months, while their A.I. revenues were only about thirty-five billion."
DeepSeek (and the like) will prevent the kind of price increases necessary for them to pay back hundreds of billions of dollars already spent, much less pay for more. If they don't find a way to make LLMs do significantly more than they do thus far, and a market willing to pay hundreds of billions of dollars for them to do it, and some kind of "moat" to prevent DeepSeek and the like from undercutting them, they will collapse under the weight of their own expenses.
mgfist 7 minutes ago [-]
DeepSeek is also undercutting itself. No one is making a profit here, everyone is trying to gobble market share. Even if you have the best model and don't care to make a dime, inference is very expensive.
disgruntledphd2 1 minutes ago [-]
I'd be surprised if Google weren't closer to profitability than basically anyone else, as they have their own hardware and have been running these kinds of applications for much longer than anyone else.
brainwipe 21 minutes ago [-]
The title is irritating, conflating AI with LLMs. LLMs are a subset of AI. I expect future systems will be mobs of expert AI agents rather than relying on LLMs to do everything. An LLM will likely be in the mix for at least the natural language processing but I wouldn't bet the farm on them alone.
DanHulton 15 minutes ago [-]
That battle was long-ago lost when the leading LLM companies and organizations insisted on referring to their products and models solely as "AI", not the more-specific "LLMs". Implementers of that technology followed suit, and that's just what it means now.
You can't blame the New Yorker for using the term in its modern, common parlance.
brookst 3 minutes ago [-]
Sure I can. If someone writing for the New Yorker has conflated the two concepts and is drawing bad conclusions because of it, that’s bad writing.
A good writer would tease apart this difference. That’s literally what good writing is about: giving a deeper understanding than a lay person would have.
dasil003 5 minutes ago [-]
Agreed, and ultimately it's fine because they're talking about products not technology. If these products go in a completely different direction and LLMs become obsolete the AI label will adapt just fine. Once these things hit common parlance there's no point in arguing technical specificity as 99.99% of the people using the term don't care, will never care, and language will follow their usage not the angry pedant.
simonw 5 minutes ago [-]
If the New Yorker published a story titled "What if LLMs Don't Get Better Than This?" I expect the portion of their readers who understood what that title meant would be pretty tiny.
DanielHB 17 minutes ago [-]
The computing power alone of all these gpus would bring a revolution in simulation software. I mean 0 AI/machine-learning, just being able to simulate much more things than we can.
Most industry-specific simulation software is REALLY crap, most from the 90s and 80s and barely evolved since then. Many stuck on single core CPUs.
bee_rider 4 minutes ago [-]
It could be a nice side-effect of having all this “LLM hardware” built into everything, nice little throughput focused accelerators in everybody’s computers.
I think if I were starting grad school now and wanted some easy points, I’d be looking at mixed precision numerical algorithms. Either coming up with new ones, or applying them in the sciences.
lokar 13 minutes ago [-]
A* search, literally textbook AI, is still doing great work.
qcnguy 40 minutes ago [-]
> You didn’t need a bar chart to recognize that GPT-4 had leaped ahead of anything that had come before.
You did though. I remember when GPT-4 was announced, OpenAI downplayed it and Altman said the difference was subtle and wouldn't be immediately apparent. For a lot of the stuff ChatGPT was being used for the gap between 3 and 4 wasn't going to really leap out at you.
In the lead up to the announcement, Altman has set the bar low by suggesting people will be disappointed and telling his Twitter followers that “we really appreciate feedback on its shortcomings.”
OpenAI described the distinction between GPT-3.5—the previous version of the technology—and GPT 4, as subtle in situations when users are having a “casual conversation” with the technology. “The difference comes out when the complexity of the task reaches a sufficient threshold—GPT-4 is more reliable, creative, and able to handle much more nuanced instructions than GPT-3.5,” a research blog post read.
In the years since we got a lot more demanding of our models. Back then people were happy if they got models to write a small simple function and it worked. Now they expect models to manipulate large production codebases and get it right first time. So, the difference between GPT-3 and GPT-4 would be more apparent. But at the time, the reaction was somewhat muted.
supriyo-biswas 30 minutes ago [-]
> Back then people were happy if they got models to write a small simple function and it worked. Now they expect models to manipulate large production codebases and get it right first time.
This push is mostly coming from the C-level and the hustler types, both of which need this to work out in order for their employeeless corporation fantasy to work out.
delusional 15 minutes ago [-]
I'm not going to say that nobody is expecting it to do these things, but I don't think they should. It's still unable to write a simple function.
What we've seen isn't a reasonable increase in expectations based upon validation of previous experiments. Instead it's racking up of expectations by all the signals of success. When they time and time again take in more VC cash at ever greater valuations, we are forced to assume they want to do something more, and since they get the cash we have to assume somebody believes them.
Its a pyramid scheme, but instead of paying out earlier investors with the later investors cash its a confidence pyramid scheme. They obsolete the previous investors valuations by making bigger claims with larger expectations. Then they use those larger expectations as proof they already fulfilled the previous expectations.
kerblang 8 minutes ago [-]
They didn't answer much of the "What if," though... Am just imagining the massive financial losses taken by so many, and if a bailout becomes necessary, because too-big-to-fail now means Microsoft, Google, Facebook et al since we transferred so much of financial engineering economics onto them since '08.
danjl 4 minutes ago [-]
Those three companies have products outside AI and won't die quickly. The ones that will collapse are betting exclusively on improvements in AI. It will be fun to watch the VC money burn.
woodpanel 3 minutes ago [-]
Last time I've checked each of these companies were still hugely profitable. So it's not going to be your average FANG in trouble here, but rather VCs and others who've jumped onto the AI-craze
zahirbmirza 8 minutes ago [-]
AI doesn't need to get better than this. It is already saving millions of hours of previously wasted human productivity. The biggest threat to these companies if their products do not improve is the local running of LLMS. That would finally justify consumers buying more memory and processor speed.
energy123 7 minutes ago [-]
How can you say progress has stalled two weeks after LLMs won gold medals at IOI and IMO?
How can you say progress has stalled without having visibility on the compute costs of gpt-5 relative to o3?
How can you say progress has stalled by referring to changes in benchmarks at the frontier over just 3.5 months?
danjl 7 minutes ago [-]
Investors are betting on growth. The public loves the hype. As a user, I already have something useful. Schadenfreude.
EcommerceFlow 14 minutes ago [-]
OpenAi has 700+ million users. Sam recently said only 7% of Plus users were using thinking (o3)!!! That means 93% of their users were using nothing but 4o!
Clearly the OpenAi leadership saw these stats and understood the main initial goal of GPT5 is to introduce this auto-router, and not go all in on intelligence for the 3-7% who care to use it.
This is a genius move IMO, and will get tons of users to flood to ChatGPT over competitors. Grok, Gemini, etc are now fighting over scraps of the top 1% while OpenAi is going after the blue ocean of users.
throwaway0123_5 7 minutes ago [-]
> Sam recently said only 7% of Plus users were using thinking (o3)
Thinking or just o3, and over what timeframe? There were a lot of days where I would just rely on o4-mini and o4-mini (high) b.c. my queries weren't that complex and I wanted to save my o3 quota and get faster responses.
> That means 93% of their users were using nothing but 4o!
Also potentially 4.1 and 4.5?
SideburnsOfDoom 12 minutes ago [-]
If they're not paying users then they're just a liability.
jcfrei 2 minutes ago [-]
monetizing those will come eventually - it's just hard to get right
Mistletoe 8 minutes ago [-]
The bear market decade the stock market has been putting off since 2021 with this AI gasping phase happens.
AI is so new and so powerful, that we don't really know how to use it yet. The next step is orchestration. LLMs are already powerful but they need to be scaled horizontally. "One shotting" something with a single call to an LLM should never be expected to work. That's not how the human brain works. We iterate, we collaborate with others, we reflect... We've already unlocked the hard and "mysterious" part, now we just need time to orchestrate and network it.
monkpit 5 minutes ago [-]
I think you’re right - even if we accept the premise that there’s only room for minor marginal improvements, there’s vast amounts of room for improvement with integrations, mcp, orchestration, prompting, etc. I’m talking mostly about coding agents here but it applies more widely.
It’s a completely new tool, it’s like inventing the internal combustion engine and then going, “well, I guess that’s it, it’s kinda neat I guess.”
varelse 7 minutes ago [-]
[dead]
fuzzfactor 2 hours ago [-]
>What If A.I. Doesn't Get Better Than This?
What if it does?
There's a certain type of fear . . .
"It's the fear . . . they're gonna take my job away . . . "
It's the fear . . . I'll be working here the rest of my days . . . "
-- David Fahl
Same fear, different day.
lenerdenator 13 minutes ago [-]
Nah, I'm not afraid of working here the rest of my days. Consistent paycheck, benefits, challenging-but-rewarding work.
If you provide people with that they typically shut up and stay out of the way. Everyone should be more afraid of the former than the latter.
izzydata 7 minutes ago [-]
There is no plan in place at all for the outcome of most work becoming redundant. At least in the US I highly doubt we will be capable of implementing some system such as UBI for the benefit of all citizens so everyone can take advantage of most work being automated. Everyone will be left to pick up scraps and barely survive.
But I am extremely skeptical that current "AI" will be capable of eliminating so much of the modern workforce any time soon if ever. I can see it becoming a common place tool, maybe it already has, but not as a human replacement.
piskov 24 minutes ago [-]
Because every s-curve looks like an exponent for those in the start.
I mean look at the first plane, then first air-jets: it’s understandable to assume we would travel the galaxy in something like 2050.
Meanwhile planes are basically the same last 60 years.
LLMs are great but I firmly believe that in 2100 all is basically the same as in 2020: no free energy (fusion), no AGI.
marviio 8 minutes ago [-]
OTOH: First flight: 1903. Moon landing: 1969. Humanity went from ”Look, we’re 3 meters off the ground!” to “We just parked on the Moon” in barely a lifetime. 66 years.
CagedCoder 4 minutes ago [-]
And how much further have we gotten past the moon since then?
This isn't an OTOH, it's just another example of looking at the exponential part of the S-curve.
SAI_Peregrinus 7 minutes ago [-]
Nature abhors an exponential. They all seem to either turn out to be sigmoid or collapse entirely.
dr-detroit 14 minutes ago [-]
[dead]
Rendered at 14:35:39 GMT+0000 (Coordinated Universal Time) with Vercel.
LLMs were not intended to be the core foundation of artificial intelligence but an experiment around deep learning and language. Its success was an almost accidental byproduct of the availability of large amount of structured data to train from and the natural human bias to be tricked by language (Eliza effect).
But human language itself is quite weak from a cognitive perspective and we end up with an extremely broad but shallow and brittle model. The recent and extremely costly attempts to build reasoning around don't seem much more promising than using a lot of hardcoded heuristics, basically ignoring the bitter lesson.
I've seen many argue that a real human level AI should be trained from real-world experience, I am not sure this is true, but training should likely start from lower-level data than language, still using tokens and huge scale, and probably deeper networks.
DeepSeek (and the like) will prevent the kind of price increases necessary for them to pay back hundreds of billions of dollars already spent, much less pay for more. If they don't find a way to make LLMs do significantly more than they do thus far, and a market willing to pay hundreds of billions of dollars for them to do it, and some kind of "moat" to prevent DeepSeek and the like from undercutting them, they will collapse under the weight of their own expenses.
You can't blame the New Yorker for using the term in its modern, common parlance.
A good writer would tease apart this difference. That’s literally what good writing is about: giving a deeper understanding than a lay person would have.
Most industry-specific simulation software is REALLY crap, most from the 90s and 80s and barely evolved since then. Many stuck on single core CPUs.
I think if I were starting grad school now and wanted some easy points, I’d be looking at mixed precision numerical algorithms. Either coming up with new ones, or applying them in the sciences.
You did though. I remember when GPT-4 was announced, OpenAI downplayed it and Altman said the difference was subtle and wouldn't be immediately apparent. For a lot of the stuff ChatGPT was being used for the gap between 3 and 4 wasn't going to really leap out at you.
https://fortune.com/2023/03/14/openai-releases-gpt-4-improve...
In the lead up to the announcement, Altman has set the bar low by suggesting people will be disappointed and telling his Twitter followers that “we really appreciate feedback on its shortcomings.”
OpenAI described the distinction between GPT-3.5—the previous version of the technology—and GPT 4, as subtle in situations when users are having a “casual conversation” with the technology. “The difference comes out when the complexity of the task reaches a sufficient threshold—GPT-4 is more reliable, creative, and able to handle much more nuanced instructions than GPT-3.5,” a research blog post read.
In the years since we got a lot more demanding of our models. Back then people were happy if they got models to write a small simple function and it worked. Now they expect models to manipulate large production codebases and get it right first time. So, the difference between GPT-3 and GPT-4 would be more apparent. But at the time, the reaction was somewhat muted.
This push is mostly coming from the C-level and the hustler types, both of which need this to work out in order for their employeeless corporation fantasy to work out.
What we've seen isn't a reasonable increase in expectations based upon validation of previous experiments. Instead it's racking up of expectations by all the signals of success. When they time and time again take in more VC cash at ever greater valuations, we are forced to assume they want to do something more, and since they get the cash we have to assume somebody believes them.
Its a pyramid scheme, but instead of paying out earlier investors with the later investors cash its a confidence pyramid scheme. They obsolete the previous investors valuations by making bigger claims with larger expectations. Then they use those larger expectations as proof they already fulfilled the previous expectations.
How can you say progress has stalled without having visibility on the compute costs of gpt-5 relative to o3?
How can you say progress has stalled by referring to changes in benchmarks at the frontier over just 3.5 months?
Clearly the OpenAi leadership saw these stats and understood the main initial goal of GPT5 is to introduce this auto-router, and not go all in on intelligence for the 3-7% who care to use it.
This is a genius move IMO, and will get tons of users to flood to ChatGPT over competitors. Grok, Gemini, etc are now fighting over scraps of the top 1% while OpenAi is going after the blue ocean of users.
Thinking or just o3, and over what timeframe? There were a lot of days where I would just rely on o4-mini and o4-mini (high) b.c. my queries weren't that complex and I wanted to save my o3 quota and get faster responses.
> That means 93% of their users were using nothing but 4o!
Also potentially 4.1 and 4.5?
https://www.currentmarketvaluation.com/models/s&p500-mean-re...
https://www.cell.com/fulltext/S0092-8674(00)80089-6
It’s a completely new tool, it’s like inventing the internal combustion engine and then going, “well, I guess that’s it, it’s kinda neat I guess.”
What if it does?
There's a certain type of fear . . .
-- David FahlSame fear, different day.
If you provide people with that they typically shut up and stay out of the way. Everyone should be more afraid of the former than the latter.
But I am extremely skeptical that current "AI" will be capable of eliminating so much of the modern workforce any time soon if ever. I can see it becoming a common place tool, maybe it already has, but not as a human replacement.
I mean look at the first plane, then first air-jets: it’s understandable to assume we would travel the galaxy in something like 2050.
Meanwhile planes are basically the same last 60 years.
LLMs are great but I firmly believe that in 2100 all is basically the same as in 2020: no free energy (fusion), no AGI.
This isn't an OTOH, it's just another example of looking at the exponential part of the S-curve.