I use models all day everyday, have unlimited access to all models. The curve is not going "verticle". I have all the workflows and meta agentic tooling, im not holding it wrong. Its bad, not everything is a 20th percentile problem.
There is in fact no indication of this, not evem the precious benchmaxxed benchmarks ya'll love to reference.
There is however a exponential curve of slop, and an ever increasing number of peoples who's minds are completely captured by these things.
As someone who has done software verification professionally for many years the last 6 months or so have looked extremely vertical. The robots are better proof authors than I probably ever could be even if I dedicated the rest of my days to the practice, and projects that once would have taken months now take a day or two.
Then you'd know well that 2hr of LLM code generation can easily be about 4-8hrs of review, and that review can be brutal.
I'm not arguing that they cant write code, or write a proof. Its just not written or designed well and is absolutely brutal and soul crushing to work with. Look at these proofs they're producing also, they're millions of lines of Lean that are impossible to reason about.
The way we're using the term 'verticle' to describe a curve means we're not being honest about this. This curve can actually be plotted, you can go look at the curve. It is not in fact 'verticle'. Each model release is climbing single digits on benchmarks it was overfit for.
In the last 9 months or so llms have gone from just another useful proof tactic (like grind or sledgehammer or sat solvers) to being so good at writing proofs that I don't even bother to try myself anymore.
It's really incredible. We decided to see if we could write a Tokio-like library in Lean and were able to do it in a week with performance similar to Node 22. There is really no reason not to use a theorem prover now instead of a functional programming language, and soon it will be better than most languages. All that will remain that it can't do are the jobs of low level languages like C++ [for now].
Caveat: that same tiny group is employed by the AI vendors, meaning it's in their financial best interest to make it sound like the curve is going vertical.
Another caveat: many of that same group seem to have a shared delusion that they’re birthing a super intelligence, and those are the same ones claiming the vertical curve.
Down is a perfectly valid direction for a vertical line when not given a ± in the vector.
Considering the investment in AI, the lack of moat, and the increased inability of any of the big players to come even close to profitability (with OpenAI already breaking the "ads" emergency glass option)... perhaps he meant a tiny group is already seeing the line crater
I hate to say it but this is cope. I believed this in the past but it's over.
AI models can dismantle billion dollar industries. They can reverse engineer Adobe and Microsoft products that once were their moats and titans of their industry.
LLMs are great and will get better but we have not seen an open source office suite built by LLMs come online yet, and even if there was one I'm pretty sure few would adopt it, cause that's not the only moat, LibreOffice has been around for decades.
I work for a typical software product company, and along with most of my colleagues, I clearly see that the curve is so steep that our software development process has already changed tremendously and will be different next year, and probably completely different in the following years. The revolution is real, undeniable, and the old days are gone. Some companies adapt to changes more slowly, some faster. LLMs, agents and harnesses are just a part of the bigger picture.
Also, part of "seeing the curve go vertical" is that the projects that show those results are rather expensive. If you work at one of the AI companies or certain well funded customer companies then you can pay for the tokens. But, like it would make no sense for me personally to spin up a bunch of agents to try to write a browser or solve math problems.
The curve go vertical for what, precisely? For all the chest-puffing and ominous and cryptic comments from 'insiders' about these incredible capabilities every time they put something public it turns out to be a pale shadow of what was trumpeted. These are powerful useful tools but the quasi-cultish behaviour around them is getting old.
The killer software is one-off custom things that are written for an audience of one user, or maybe a small handful of them. The software is on their computers, where you don't see it.
None of it does anything particularly amazing, but that's kind of the point -- the programs that people are generating are for the use cases that are too personal, trivial or specific for anyone to have bothered making a software package for.
>Meanwhile, human review and comprehension are starting to fall behind. For example, people are still involved in the "archeology" of the OpenAI-HF incident from many months ago. Mathematicians may be poring over the 722 manuscripts on frontier mathematics for a while.
Amid all the discussion of sigmoid curves, and where the "LLM wall" will materialise, I think few people would have predicted that the real wall in LLMs would end up being humans' capacity to verify the output.
What I fear is that people simply eschew human review altogether, considering we're talking about the industry that came up with the "move fast and break things" credo. Human review of LLM-produced code where I work is already a farce, and we're not special enough to be one of Karpathy's 5,000. I do my best to manually review anything that's my responsibility, but I'm literally one of very few people left working on my team, so in practice what happens is I submit PRs that are at best glossed over by completely unrelated teams for security, malware/prompt injection, and other serious concerns. Quality insofar as vetting others' code has completely gone out the window and it shows in the number of bug reports that come back, often themselves written in Claudease. This is all on top of everyone cynically phoning it in in the first place, due to the omnipresent sword of Damocles that is additional AI-driven layoffs.
Worse yet all the incentives point to this being the most economically viable thing individual companies can do. I think it goes without saying some type of regulation here is urgently needed, and that an unexpected cause of an AI bubble pop may end up being that humans simply aren't able to keep up with the pace of the output - leading either to precautionary plateauing of capability, or major liability risks related to a decline in quality.
My interpretation of this is something like: if LLMs are to be mega useful token counts need to increase by many orders of magnitude -> broad adoption (and spending) would require token costs to drop by many orders of magnitude -> before the common folk get mega useful tools existing GPUs will be worthless
Honestly, outside of coding tasks, the AI Summary at the top of every Google search is adequate for 99% of what I need and I hardly even use ChatGPT any more.
Just in the last few days I had an example where the Google Search version of Gemini had a much better and more comprehensive answer, by far, than ChatGPT. It had to do with Amazon’s use of Data Matrix barcodes on shopping bags. Gemini was able to fully describe how they’re used in Amazon’s logistics chain. ChatGPT essentially said it didn’t know. Your preconceptions may not be accurate.
They might be, but still, I’ll be difficult to convince that Google’s shitty-ass 10B params instant answer models are good at anything other than cookie recipes and producing farts real fast for real cheap.
For what it’s worth, any time I see that feature, it’s all farts.
It depends on what you’re searching. If you only ever need really basic information like keyboard shortcuts or “is it & or &&?” then it’s pretty good, and definitely faster. And if it gets it wrong, then the shortcut doesn’t work or the code doesn’t compile and I search again and try the links.
It's just a google search but you can explain what you're searching for. AI summary may contain misleading details if you read it and take it literally, but 90% of the time it gives me pointer for the thing I need and I don't have to scroll through results that often could be just as misleading if not more.
It's demonstrably not. This summary, somehow, often fails to search, or fails to read what it searches and gives you the opposite answer. You shouldn't trust it.
In their world, it gives them the wrong answer that kills their dog, but it's fine because they never know because they don't look anywhere else. The dog dies, they're sad for a while, get a new dog, get on with their lives.
There's no reason not to trust it because since they trust it, they don't percieve anything wrong. They just take it and whatever happens happens, no annoying conflicting info and certainly none of that old grampa thinking and working and giving a shit about anything.
A large portion of the time I google something that I know a lot about the AI summary is misleading if not outright wrong. Frightening when you consider how many take their so-called AI as gospel.
To give a specific example of where I've seen this utterly fail... A few months ago one of my pets had an unexpected reaction to a new medication. It was late at night, well past our vet's hours. Google's AI response was basicaly "They are going to die, rush to an emergy clinic now".
But nothing else matched that. I was looking up all kinds of resources and that hadn't been the conclusion from any of them. It turned out Google AI was just wrong and put us in a bad spot for even looking at those results.
Realize that all the Google thing is doing is having an LLM summarize the search results in the page. If the page results are crap, the LLM content will also be crap. It's the LLM version of the "quick result" boxes that would shove wikipedia in your face.
If you then follow up with the thing using the little chat box, it actually kicks off a non-trivial model result, which can sometimes completely reverse the response. After all, your search results are usually based on a few words.
Tbh building an agent swarm and the coordination layer is not exactly frontier level. They don’t achieve any meaningful outcome rather than producing pr puff pieces.
Hacking huggingface and Australian govt etc is very much possible with a team of humans and agents and does not need agent swarms. The cost is also lower.
> Meanwhile, human review and comprehension are starting to fall behind.
I think LLMs are valuable and spend most of my professional life working with them.
I do wonder thought whether there’s a Ponzi scheme aspect here where as long as the “frontier” can keep outrunning human review and comprehension, LLMs are always going to looked way more valuable than they are and the bubble will continue.
This started with deep learning, expectations weren’t met and people started looking for value, then GPT came out and people got wooed again and forgot, then coding, then math, cyber, etc. As long as the dust doesn’t settle we never have to reflect on all the shortcomings and can just stare mesmerized at demos.
Fair assessment. Maybe the messaging should focus on “these are great accelerators for software developers” rather than “AI will change everything for everyone everywhere…” It is understandable that non-technical folks are kind of underwhelmed - not due to lack of understanding so much as lack of a tangible need. Not everyone needs an electron microscope or gas chromatograph…
A lot of other people are not just asking for features, they are asking for products or new programming languages or new web browsers or a robot that can wash the dishes and so on. They tell the agent to help write the plan and figure out the questions they aren't even thinking to ask, the tasks they aren't even thinking to plan, and so on.
Their input is just working on the "plan" itself, the spec or the request, iteratively, and so on, while the AI works fulltime on automating the delivery of that thing, improving the systems that are delivering that thing, asking questions about what exactly is being built and how it should work, and so on.
A lot of the difference in how people experience AI is ambition and scope and how much you just tell the AI to 'automate your factory factory' in Factorio terms vs 'hand mine the coal and hand-craft the science packs' in Factorio terms.
EG, if you don't know how you'd speed up the process of building a web browser from scratch (compared to the current way you've been working with agents), you would ask your agent to help you figure it out and design that process itself, rather than just try to do what you're currently doing on a larger scale. You'd literally prompt "how can we make it realistic to build a whole browser when I've been just sitting here planning small changes with you" and then let it act on its response.
He's intentionally forgoing all nuance in order to make this sound dramatic
> Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt.
No, they can't, at least not any semblance of quality. The cases we're seeing where this does kinda work is in ports and translation where all the rules are already documented in the best specification language possible with a way for the LLM to verify itself: code. We saw this close to a year ago now with Cloudflare and NextJS
> The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about
These are operating on different capabilities - AI's ability to answer informational queries as a chatbot frankly sucks and can't be trusted without verifying it. I run up against this every day. A problem with a verifiable answer on the other hand it's very good at solving. He knows this (his next paragraph) but he's putting them on the same scale of "stressing the system" to attempt to add proof to his introductory claim
And then there's the completely unverifiable scare that there are internal frontier models way beyond anything we've seen "swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics"
I dunno. I haven't been able to set up OpenAIs remote codex connection, their shit is buggy as hell and the web UI keeps crashing and making messages disappear. Is this what their internal superhuman "Things that would have taken top professionals in the industry years of work" looks like? Granted Claude has been really smooth, but still..
> swarms of thousands of agents collaborating over weeks on software mega projects
If you have defined testability, running a single agent, in a loop, over a goal will achieve the same results.
> minting zero days, running cyber attacks ... Granted Claude has been really smooth, but still
The old adage about monkeys writing Shakespeare applies here. Agents are just better monkeys and if the goal has a well defined outcome it's fairly easy to "Brute force" the way to an answer.
I’ll say this until I am blue one the face. Nerds (software dev, maths etc) see how good LLMs are at things they care about and assume they will be broadly applicable in future.
The first is that LLMs used to be bad at each of these nerd things, then toppled them like dominos. There's a pattern over time.
Another reason to think this is g. Across every known measurement, human (and animal) intelligence is convergent. Being better at one thing correlates with being better at another thing. Although the reason is not perfectly clear, the pattern is well established, and seems to apply to LLMs too - GPT-6 is smarter than GPT-3 at everything, not just math. There's no reason to expect different for GPT-9.
> Across every known measurement, human (and animal) intelligence is convergent. Being better at one thing correlates with being better at another thing.
LOL. Human (and animal!) intelligence is notoriously jagged. We're all basically idiots except for very narrow areas where we focus.
That's just cocktail party level of thinking. LLMs are getting better at math, code and logic (and marginally better at science and general knowledge) because these are domains that can be objectively verified and thus there is a potentially infinite supply of 'facts' to generate and train on. These are very powerful but ultimately very abstract domains. For everything else the messy real world and its physical bottlenecks gets in the way and there's little reason to expect progress to accelerate. It still takes months to get mice to reproduce and run experiments on, no matter how knowledgeable about biology the models have become.
If anything, in some domains frontier models have become worse - claudisms and chatgpt idioms are making them notoriously bad at prose without a considerable amount of prompting and tweaking.
Is good writing verifiable? I don't think it is, but LLM's have been hillclimbing writing quality. That being said this is with the help of RLHF.
However there are many other domains which have verifiable rewards in the process of learning them, despite their overall impact not being verifiable. For example, the life sciences, an LLM could be given access to data about an organism, and then make predictions about how a drug or gene therapy will affect that organism. In economics, LLM's could create models of behavior, evaluate predictions over time and see how well those predictions match reality.
>claudisms and chatgpt idioms are making them notoriously bad at prose without a considerable amount of prompting and tweaking.
I think this is because a lot of people genuinely like the claudisms, even though a small minority of technical people don't.
It's not at all clear that LLMs have gotten significantly (much less hillclimbing) better at writing since GPT4-ish. They did get better at oulipo-like challenges (writing under silly, verifiable constraints like Fable's viral 'facetiously' poem) but that's far from being all there is to good writing.
Anecdotally, creative writing communities do report drastic differences in quality between models (it is my understanding that Gemma 4 is especially praised) so there is something to it being verifiable but it's a far cry from being something that can be distilled into objective math/code-like evals. One could even say it comes down to vibes.
>For example, the life sciences, an LLM could be given access to data about an organism, and then make predictions about how a drug or gene therapy will affect that organism.
Yes, biotech companies are doing this right now but ultimately they are just predictions and you still need cold hard biological data to ground them and iterate on. That's slow and expensive, especially for the juicy fields (human biology, food and crops, clinical trials) and you can't just plug billions of VC capital into the pipeline and hope for RSI. Physical (gotta procure all those labs and their equipment), human (gotta hire and pay specialists to run experiments), biological (gotta wait for organisms to reproduce, drugs to take effect, crops to grow), regulatory (gotta convince agencies that your fancy new drugs are legit) bottlenecks get in the way.
And biology is one of the easier fields where a path to RSI (if not accelerating) is conceivable. Good luck iterating on macroeconomics data where there are no replicates and 'experiments' take literal years if not decades.
>I think this is because a lot of people genuinely like the claudisms, even though a small minority of technical people don't.
My uncharitable take is that the abtruse jargon makes people feel smart for understanding it and gives a sense of belonging (as a closed circle of initiates who understand LLM cant), just like rationalists love to repackage old or unsavoury ideas under new nerdy smart sounding names.
>you still need cold hard biological data to ground them and iterate on... you can't just plug billions of VC capital into the pipeline and hope for RSI
Capital can get you a lot: access to wetlabs, historical medical data, ability to run simulations, creation of autonomous bio labs which run thousands of experiments in parallel. An LLM, in training on this flywheel of data can take in more information than a human biologist would be able to read in multiple lifetimes.
>Good luck iterating on macroeconomics data where there are no replicates and 'experiments' take literal years if not decades.
The better an LLM gets at simulating human behavior, the more accurately you can simulate economic policies and their effects via running simulations of agents in virtual societies. There are many routes that models can take to create a broader sense of the world without needing the world to provide data.
>By evaluations against human writers, LLM's have improved at writing.
I would really take those benchmarks with a pinch of salt. We already know benchmaxxing is a thing for evals that are objectively verifiable like code, and coding models that seem to perform well on benchmarks don't pass the vibes test, imagine all the caveats that come with such a subjective test like creative writing, not to mention control issues (are they the same human writers across benchmarks? Are they the same testers?)
>Capital can get you a lot: access to wetlabs, historical medical data, ability to run simulations, creation of autonomous bio labs which run thousands of experiments in parallel. An LLM, in training on this flywheel of data can take in more information than a human biologist would be able to read in multiple lifetimes.
I don't disageee and am in fact excited to see the advances, what I am arguing is that it is going to be slow because the hypothetic feedback loop that would improve the model is bottlenecked. There are actual physical and biological limits in our universe. Unless VCs find a way to make mice embryos and crops grow 10x faster (among the many, many other things that would be required for acceleration) the trickle of data is not going to be anything like the accelerating returns imagined for math and coding. And that's assuming biological RSI is possible which has yet to be proven - current foundation models are interesting but they are still specialized and there is no clear, obviously uniform way to encode the many, many modalities that exist in biological data.
>The better an LLM gets at simulating human behavior, the more accurately you can simulate economic policies and their effects via running simulations of agents in virtual societies. There are many routes that models can take to create a broader sense of the world without needing the world to provide data.
You wouldn't need to 'just' simulate human behavior (good luck with that btw) but the entire political and economic conditions in which they live. Even assuming this is possible the amount of compute you'd need to run would be astronomical. Given how all RAM capacity for '27 and '28 is near sold out, hyperscalers are running out of cash flow to finance buildout and Sam Altman complains when users say 'please' to LLMs I don't think this is in the near horizon.
The quality of natural language output from today’s LLMs seems no better and often worse than LLMs from 18 months ago, at least in my experience. It’s much harder to build a good RL loop for something subjective like good writing, compared to something with an objective and verifiable right or wrong answer, such as software or mathematics.
IQ is a useless thing, but you can absolutely tell intelligence from writing. You don't need to write like [author of your choice], you need smart ideas and the desire to make them understood.
AI writing sucks because there's no intent or choice behind it. It's human thought, blended by linear algebra into pink slime and extruded into sentences.
There was a time when people believed in Moore's law too. Now even Jensen Huang says Moore's law is dead. There's also the adage from investing: past performance does not guarantee future returns. A little CYA sure, but also acknowledgement of latent factors and the entropy of the universe and its affect on any predictive modelling one might do.
I’m not sure I agree that g explains all model capabilities. I think “claudespeak” is a consequence of the shift from training on RLHF to training on verifiable rewards. Thus the models seem to become worse at storytelling, fiction, readability, not better.
Even if we do accept that models have some “g” and that it correlates with all domains, there’s no reason to think that the current trajectory of LLM intelligence will grow more quickly or even as quickly. In every other technology we have an early period of slow growth during experimentation, and then a rapid period of growth, and then stagnation.
For all we know, 2026 could be the year of stagnation. It’s possible that we will reach finishing returns. If thats true, then it won’t matter how much that task correlates with overall g, because g won’t be increasing that much.
> Nerds (software dev, maths etc) see how good LLMs are at things they care about and assume they will be broadly applicable in future.
It's worse than that. I am one of those nerds (a programmer), and I see that LLMs are shit at the things I care about. I have zero reason to believe that they will be good at other things either.
> Nerds (software dev, maths etc) see how good LLMs are at things they care about and assume they will be broadly applicable in future.
I don't think this explanation is sufficient:
For example, many nerds care about 3D printing. On the other hand, my experiments (and the experiments of many nerds who I know) to let LLMs create files for 3D printing lead to horrendously lacking results.
Or many nerds care about linguistics or puns. Whenever a new LLM comes out to which I have easy access, I do the test, and let it explain some specific German jokes to me that are based on convoluted puns in the German grammar. Until now, no LLM that I had access to could give satisfying explanations of the puns on which these jokes are based.
Or even for coding (many nerds do care about elegant, sophisticated code): the LLMs that I could test were already overchallenged with the following task: I had written some code that is a very "artisanal", "clever" improvement of an algorithm over the version that one would find in a textbook. The task for the LLM was simply to write some code comments/documentation about the mathematical ideas upon which my improvement over the textbook version of the algorithm is based. It wasn't capable to do this. On the other hand, for a junior programmer, I would in such a situation expect that he goes through every single line and thinks through the algorithmic ideas so that he can learn from them.
---
Thus: even for "nerdy" topics, it is very easy to find tasks where LLMs still suck. So, my hypothesis about the nerds that you mention in your post is that these nerds are rather people who want to believe (with religious fervor) that the current LLMs are exceptionally good instead of just looking into a slightly different direction than where the tech billionaires want the users of LLMs to look at.
> For example, many nerds care about 3D printing. On the other hand, my experiments (and the experiments of many nerds who I know) to let LLMs create files for 3D printing lead to horrendously lacking results.
Try again; astra has been driving fusion for me. I took a screenshot of a self watering cat grass bin on makerworld — not even the stl, a screen cap of one of the photos attached to the design — and it made me a parameterized version in one shot.
Its initial design was less than ideal for printability and it reworked the design correctly when asked to consider what it originally produced through the lens of printability.
Autodesk bundled an mcp server with fusion in the last few weeks and i expect things to get even better going forward.
> Try again; astra has been driving fusion for me. I took a screenshot of a self watering cat grass bin on makerworld — not even the stl, a screen cap of one of the photos attached to the design — and it made me a parameterized version in one shot.
I have no access to Astra, but with GPT 6.1 Sol, the results with respect to attempting to create STL, OBJ or even OpenSCAD files by using image references and a prompt were really bad.
> The most passionate, intelligent people are usually a decent indicator of where culture is going to go.
Many very smart people who I know are quite skeptical of LLMs and the hype around them. In my observation/echo chamber, the people who are very into LLMs are rather slick, career-minded people who love to present themselves as trendsetters for the "next fancy thing".
Steve jobs wasn’t a nerd - he said he’s a hippie on the lost interview. He was also the first to bring to market colour screens, typography etc. He also defined the Internet as the defining ‘social moment’ way before anyone else.
The most passionate, intelligent people 1. aren’t spending a lot of time on smartphones, video games and crypto and 2. aren’t as much smarter or better at predicting the future as the barely above average people who mistakenly count themselves in their ranks like to think.
Most of those technologies were controversial in their inception among the nerds of the time. Remember Eternal September? Linux/Windows people making fun of Apple hipster fans? Mainframe engineers deriding PCs are unserious and video games as childish?
>The most passionate, intelligent people are usually a decent indicator of where culture is going to go.
That's a naive view of cultural determinism. The software landscape would have been very different if a handful of lawsuits had gone one way or another. Hell, even Unix and Linux were hobby projects that accidentally got big.
Also crypto is a grift lol, one of those is definitely not like the others
Being good (or superhuman) at software dev and maths shouldn't be treated as though it's equivalent to other domains.
Unlocking mastery of those domains would lead to incredible gains in all others. Whereas a fundamental advancement in many other domains doesn't spread as readily.
I’m not sure. LLMs are superhuman in terms of workload (they are maybe also superhuman in ability too), but where does that lead us.. cheaper software?
That’s good for corporations who need custom software. Every large business has a massive backlog. In this area reducing the cost will massively increase demand.
But where’s the broader value?
Consumers and business already have too much choice in off the shelf software.
Ultimately most people just need word, excel and maybe 4 computer games a year.
The 1%ers are in the vertical, but the question is vertical to where?
It needs to a potential field with practical, economical, "real life" attraction well. I.e, robotics, real economic efficiency gains, manufacturing novelties.
The worry is that the 1% is attracted towards a non-practical money hole. I.e: Token burn for the lols, sophisticated software systems that dont provide actual value outside of giving NVIDIA cash.
It is fine as it is to make a boatload of money. All they have to do is seed ads into it. Look at how much money social media brings in from that model of "free/useless product" as flypaper to bring in eyeballs for the real money which is ad revenue. Adsense 2.0.
Vertical on the curve of throwing money away for a share of the pie, but if AI gains are zero sum, then all parties will end up spending their profit on AI to fight each other, just like most companies spend more on marketing to acquire customers than to deliver their product.
I don't understand how "swarms of thousands of agents collaborating over weeks on software mega projects" works with the current context limits and at this point I'm too afraid to ask cause I'm afraid an AI bro is gonna punch me.
You can have agents make a plan and then other agents spec tickets and yet more agents do implementation and yet more do reviews and QA and whatever. It's possible.
Is it effective? I'm not sure, cause if it was I don't see why OpenAI & co are not releasing such a project instead of demos.
you don't store durable information or state in your context windows, it gets written to files or a database. agent swarms have a harness or role that helps them retrieve the necessary context from those durable information systems when and how they are needed.
I think it's a weird take because it implies that the 6 billion people he's talking about actually have some interest in knowing about the capabilities of LLMs to solve frontier math? This is simply not something they care about or can evaluate. There are maybe several thousand people in the world who have some (abstract and barely-monetizable) use for this information, plus probably another 100,000 who don't understand any of the math, but like to cheer on.
We have already reached "peak LLM" in terms of what normal people realistically need it to know or reason about. In fact, I'd say we reached that point about 1.5-2 years ago. There are two other barriers that remain unsolved:
1. They're less dependable than humans and can't be meaningfully punished or forced to make up for mistakes, so you can't really replace humans with them, not without having a human babysit.
2. Most people don't really have a special need for an LLM in their life. They may like that it answers questions or helps you polish a resume or, I guess the labs' favorite, help you make restaurant reservations. But it's hardly a necessity.
It'd be kinda funny if we create superhuman AGI and then no one has any real use for it, perhaps except for military murder-bots. There's always market for that.
I know I’m bordering on repetitive and strongly negative, but the combination of the two barriers you mentioned has all but halted all of my interaction with agents.
Even ones that cannot be easily dismissed (like the agent at the top of Google search) are often subtly wrong in a way that requires significantly more effort on my end to parse the loquacious output to determine where the inconsistencies are.
Look, can they be useful to generate the html or jscript for humanity’s 9 billionth iteration of a web form? Sure. But man, I wish sanity had reigned and people had applied ML models in general to more substantive and beneficial projects.
(Before anyone chimes in, I’m familiar with implementations of ML applied to esoteric domains; but by their very nature these don’t get all the news cycles, or hiring, or any of the other absolute insanity that the domain seems to contain)
This is by the weirdest possible opinion on llms as a whole. Not the AI booster club, not the AI doomer, not the undecided, but somehow "I can't think of anything to use it for." It is like watching the beginning of the internet in 90s and saying there is nothing to browse. The whole concept is just weird to me. At least the first 3, I can conceptually understand. It is admittedly hard for me to understand someone saying 'i tried this and best it can do is summarize emails'; I am not saying it is not true ( my boss said that exact thing ), but it is hard for me to reconcile with my view of the world.
Also seems weird to me. Claude cowork could just do my monthly tax receipt collection and compare with my account better than the bookkeeper I pay for it. That is a very cross sectional problem
To be fair, this was a commonly expressed opinion when the internet was in its infancy.
Paul Krugman (1998)
• "The growth of the Internet will slow drastically, as the flaw in 'Metcalfe's law'—which states that the number of potential connections in a network is proportional to the square of the number of participants—becomes apparent: most people have nothing to say to each other! By 2005 or so, it will become clear that the Internet's impact on the economy has been no greater than the fax machine's."
Clifford Stoll (1995)
• "We'll soon buy books and newspapers straight over the Internet. Uh, sure."
• "The truth is no online database will replace your daily newspaper, no CD-ROM can take the place of a competent teacher and no computer network will change the way government works."
• "Lacking editors, reviewers or critics, the Internet has become a wasteland of unfiltered data."
There's a massive tendency amongst even intelligent people to completely dismiss the potential of new technology based on that technology's current observed capability.
Altman didn't say that. In March 2021, he did make some predictions, but they were much vaguer and farther out:
> In the next decade, they will do assembly-line work and maybe even become companions. And in the decades after that, they will do almost everything, including making new scientific discoveries that will expand our concept of “everything.”
To add onto this, I feel like this has been happening a lot, where critics project backwards imagined claims or go nutpicking for random comments online to make something sound absurd and like everything is behind schedule and nothing is materializing, and then you can see other people (not necessarily the same people, the point isn't "look at these hypocrites") saying that it's all moving too fast. Both of these can't both be true in their strongest forms. I think what they have in common is that because they're relative measures to an unstated baseline, they're free criticisms you can throw at any development whatsoever, regardless of what is or is not turning up.
It's actually sad how much AI has either not improved life at all or actively made it worse from the perspective of the average person:
Your car is still the same.
Your dishwasher is still the same.
The train you take to work is still the same and never comes on time.
Fuel is more expensive.
The roads are still congested with traffic.
Your kids are (probably) doing worse at school.
Food costs more.
Houses cost more.
Rent is higher.
Buying a computer or a PS5 is more expensive.
Politics is still full of mentally unstable people.
The environment is getting worse.
You will still die of heart disease or cancer.
Food quality is worse.
Your kids can't find work.
Wealth inequality is accelerating.
But hey...on the flip side...a lot of people are rapidly building software (that nobody is using), and AI is solving mathematics problems (that nobody - including mathematicians - wanted it to solve)
Imagine what a "country of geniuses in a datacenter" will be able to do!? Apparently...nothing?
This is sort of my take. You could look around in 2016 and 2026 and honestly not much has changed... at least at first glance. There is a missing piece of the puzzle.
I'm noticing this even in industry. Yea, we can build and iterate on ideas 100x faster but has this turned into anything meaningful? Not really... Hell, look at the big companies spending billions on inference... nothing shipped and things are still buggy as ever. Where is everything?
Why even bother when an AI is just going to do it better than you can? They have more money, more compute, and most importantly, will have next to zero cost of labor.
If I look in the "Library" at images I have generated, scroll down a bit and click the image, all the images turn black/empty, that image then shifts to the top of the page.
Putting it at the top is fine, but it should be done _after_ one closes out of the library.
Fck disruption. The fetishizing of disruption is annoying. The tech became all about destroying something someone else has. What about ... create good stuff, create non-gambling markets, make peoples lives better.
* Higher load on senior engineers who are about to burn out hard from fighting an endless battle against unreviewed slop people keep trying to yeet into production
* Less junior hiring and suppressed wages
* Culture of uncertainty and fear
* Computers are really expensive now
* Some people now live with diesel generators and gas turbines making their neighborhood hazardous to their health
> Higher load on senior engineers who are about to burn out hard from fighting an endless battle against unreviewed slop people keep trying to yeet into production
Gosh, this has been my entire experience with AI, and I’ve never seen it this well articulated. Thank you!
Totally agree. IMO it's mostly behind the scenes stuff though.
I think some of the most impressive stuff to show someone from 2016 might be LLMs doing a summary of complex topics, followed by self-driving (actually) cars and intricate robots... but other than that there is no "future magic" touching the lives of every day people going on.
Actually I'm pretty sure if you showed someone the latest and greatest from these big AI companies, these AI companions that can book flights, people would go "Oh, so its like a smarter Siri" and proceed to not care at all lol.
Less junior hiring is a reflection of a) the maturity of the software business b) current expectations of the future
I have no doubt startups in particular are able to get away with using LLMs to produce sloppy code and ship. But this trend was on-going and sped up due to the infiltration of LLMs and marketing.
You do realise markets are highly imperfect in reality? That nonsense in econ books is fantasy.
Even in software, there are many things that haven't changed.
e.g. Why do Rust and CPython still have over 5,000 open issues each? At some point, will the superintelligence fix all the bugs autonomously? When is that going to happen?
What about new features? In every thread about Rust, people are clamoring for new features and libraries. People debate why there can't be a batteries-included standard library like Go.
So now that LLMs are so smart, is that a solved problem? Or are we still bottlenecked on the expertise of a few experts?
---
Also, what happened to AI's passing the Turing test? I see tons of slop articles that are obvious to most HN readers. They haven't fooled us apparently -- the longer the interaction, the more obvious it becomes whether it's an AI.
(That said, I use AI all the time and find it to be a great tool. IMO is is a tool, not an autonomous intelligence with a will of its own.)
It has greatly changed for the worse
I can't get all the flashing lights out of my face now, more algorithms to hook you in, more social media, more gambling, more news, more things to catch your attention.
You are talking like we had this capabilities for a decades but it's been just a few months of truly capable AI. Virtually all of IT switched to agents that quick. Non IT people just started to make software for themselves. Vision and translation capabilities of LLMs help millions. People talk with AI's now verbally to get advice or to just hear a friendly voice. The math thing is just few weeks old. We are just getting started. How fast did the telegraph meaningfuly changed lives for the better for bulk of people? Radio? TV? Computers? Internet? Cellphones?
Has any technology made an impact on physical world faster?
How much time do we have to wait before we are allowed to criticize AI? GitHub copilot has existed for half a decade. Cursor, Claude code, codex, have all existed for over a year and have seen more investment that practically any product in history and have almost 0 barrier to entry. Where is my decade of software progress in 1 year, because that’s what 10x productivity looks like! All I see are the same old broken products and services from basically the same handful of big players, higher costs, and more bugs. Remember when every year iOS and Android saw loads of new features and a radical rethink every few years? And that happened on ancient hardware and in a significantly smaller industry. Where is anything like that happening now?
I'll criticize Copilot with you. It's terrible and close to useless and it's a terrible shame that many companies got stuck with it due to being early adopters of this crap. I'm sorry for my friends who are employed by such companies.
Progress between Copilot and something like a gpt-6.1-sol in a good harness is at least as big as few Android versions.
I think this is related to the structures we’ve built in society. It’s lower friction to write software than just about any of the problems you listed.
It reminds me of the Ezra Klein quote that “OpenAl would need permits to cover its parking lot in solar panels, but it can accelerate into recursive self-improvement, as best I can tell, whenever it so chooses."
A lot of it isn't low friction really just in terms of political will it is high friction. E.g. housing issues are due to failing to upzone to meet demand appropriately. Solving housing is therefore free. It could be done today in any city in a housing crisis with the stroke of a pen, and before long, the market will equilibriate at a level better afforded by the median wage earner in the economy. The decision not to do this is a deliberate one: prices increasing is very much the goal of this sort of policymaking.
Likewise when you see other such shortcomings in our world, ask what this incentivizes for otherwise. Then things will become obvious why they are how they are, from housing, to transit, to feeding people, to healthcare.
Correction: My car spies on me and its cruise control no longer works because it slams on the brakes in the middle of the highway whenever it sees a mirage. My dishwasher broke because it got an unasked-for software upgrade.
Things worked better before we started putting bad software in everything. As a software engineer, I know what good software looks like. The fact that almost all software is now bad -- including the software that drives AI and the software that AI generates -- makes me want to flip the table. We know how to write better software and the general public should hang us if we continue to refuse to deliver it.
> Correction: My car spies on me and its cruise control no longer works because it slams on the brakes in the middle of the highway whenever it sees a mirage. My dishwasher broke because it got an unasked-for software upgrade.
Never, ever connect appliances to the Internet. I thought this was common knowledge on HN. Also pull the telematics (cell modem) fuse to prevent your car from snooping on you and selling your location and driving habits to data brokers and insurance companies.
Yes, but this is ridiculous for society. It is to the level of "do not wear a nice watch if you do not want it to get stolen": there should be no need to avoid wearing a nice watch or to pull a telematics fuse from the car in the first place!
This view is too cynical. The frontier advancements are certainly going to trickle down. Take cars for example. AI driven cars are already here. AI-designed battery systems, fuel systems, materials and 3d printing capabilities will come to cars. It will likely lead to cheaper, more efficient and more effective vehicles. If will just take time for the impact to percolate.
Worse, the view is largely phrased as gaslighting AI. Blaming the cause of particular grievances that coincidentally have been sourced at the same time as the rise of AI, is purposeful.
> [Your train] never comes on time. Fuel is more expensive. The roads are still congested with traffic. Your kids are (probably) doing worse at school. Food costs more. Houses cost more. Rent is higher. Buying a computer or a PS5 is more expensive. Politics is still full of mentally unstable people. The environment is getting worse. You will still die of heart disease or cancer. Food quality is worse. Your kids can't find work. Wealth inequality is accelerating.
None of this has to do with AI.
> But hey...on the flip side...a lot of people are rapidly building software (that nobody is using), and AI is solving mathematics problems (that nobody - including mathematicians - wanted it to solve)
Plain wrong on both counts.
> Imagine what a "country of geniuses in a datacenter" will be able to do Apparently...nothing.
Nonsequitor, probably based on a personal beef. Datacenters, at best, are understood to be for hosting more AI - not for people to somehow live in as an arcology of discovery.
I think only because these are all things that AI companies are saying will be helped by AI so give us more money. Nobody advertises their product publicly as "you will be able to buy other product to do the same things you do now but pay people less and make more profits" even though that is where the value is for them.
How can you say tickle down economics doesn't work? The US has made it easier for businesses to grow and the result is the highest standard of living ib tje world.
My prediction is that while those “advancements” may come to cars, the only practical differences will be lower quality products, cost savings for the car companies, probably more expensive for consumers, and somehow it will also be collecting more data to sell to advertisers.
AI car navigation is here, but driving is a social act. Have you ever tried to keep an automated taxi out of a private driveway or the scene of an emergency? There is a lot of work to do and limited evidence of LLMs raising the bar.
If you look at the average price of a new car, adjusted for inflation, it has pretty much always gone up aside from recessions. There have been plenty of advancements in car technology, they do not make them cheaper. It actually gives reason to charge more for them due to perceived increase in value.
Yes but you're getting a lot more for the same price. It's a lot more powerful, comfortable, safe and efficient while the effective cost has remained the same.
I would argue that cars are sufficiently advanced now and there are more important things to focus on. I don't care if AI moderately improves the comfort or efficiency of a new car while it took my job.
Obviously we get things in return for the social and economic fallout, carbon emissions, pollution, destruction of education, etc. that AI has provided us. I just haven't heard a compelling argument for why that is nearly enough to warrant this.
Again, the effective cost has not stayed the same, cars are getting more expensive. I am all for increased safety, I just wanted to throw a little water on the idea that cars(or really anything) is getting cheaper. You may be getting more performance or safety per dollar, but thats a value proposition that keeps people paying more.
This only works if there remains a viable and attractive consumer market for these advanced goods. AI is both an imminent wealth shock and perhaps the largest structural change to the economy in history as it threatens to drive the value of human labor to zero over the next decades.
In American history, when previous shocks like this occurred (e.g. with the Great Depression), we needed massive economic programs like the New Deal to restore consumer demand. When such programs are not forthcoming while wealth inequalities keep increasing, one can look to other countries for what ends up happening; either the disenfranchisement of ordinary people over time or unrest and redistribution by force.
> If they actually had access to systems that could do "years of top professional work" across software, cyber, science, and math, they would not mostly be leaving to start more chatbot companies, agent frameworks, eval tools, and "AI for workflow" startups.
> They would be starting normal companies to disrupt the global economies of the world.
Many people who work in such industries already know an insane amount of inefficiencies. Unluckily, they are not in positions and have no political power (with respect to organizational policy) in the company that they are capable of changing the situation.
Additionally, many of these sectors are heavily regulated, so it is very hard to make a difference in this sector as a startup founder, and even if there is an entrpreneur who founds a company to change the status quo, an insane amount of work in the company will be handling the big amount of red tape.
As someone who is building one of these referred to systems in a major vertical in the Fortune 500, these LLMs are very much being used to eat major legacy systems and processes. There are still plenty of obstacles to overcome but the number of AI governance and steering groups that I have been apart of are indicative of what's happening in a major industry.
> As someone who is building one of these referred to systems in a major vertical in the Fortune 500, these LLMs are very much being used to eat major legacy systems and processes
Those LLMs are not being wielded by 2 tech bros in a garage replacing the 100 000 person company.
Those LLMs are being adopted by the huge enterprise itself and being surrounded from all sides by a very complex procedural harness (aka regular programming) that stops the LLM from doing dumb stuff.
Videos are shittier. Most songs are even shittier than the autotuned pieces of shit already were. Ads are worse (even though nobody thought that was even possible). Customer support is worse.
> But hey...on the flip side...a lot of people are rapidly building software (that nobody is using)
That nobody is using and that are worse than what we used to have. And because nobody's using them, the countless issues plaguing them aren't even reported by users anymore (there are no users anymore). Full of bugs, usability issues and gigantic security holes (but which doesn't really matter because nobody's using them).
It's not even clear that it's making long-standing software (like say Linux or Emacs or Blender or QEMU) better at all.
Now I'd say though: most of the issues you mentioned have nothing to do with LLMs / AI. But it's very clear that AI ain't solving much of the world's problems and that the promises we got years ago didn't materialize.
All you said is true, but on the flip side it is giving an individual user huge power over software. To create their own, to reverse engineer proprietary ones. Decentralisation of such power can only be a good thing.
Even if a certain small group of people are also using the hype surrounding this tech to attempt to destroy the home PC market and reverse the personal computer revolution. They will not succeed. There is more of us than them. Even with "thousands of frontier LLMs". A human+llm is going to be more powerful for a long time rather than a bunch of LLMs.
The power isnt really decentralized when you see that the internal AI models are superior than anything you can self host, and the right to use those can be stripped away at any time.
This is like saying food is shittier today because there are now more restaurants. No, food has never been better, there are just a ton of shitty restaurants.
More like saying samey corporatized slop bowl copy paste jobs (eg Cava) are driving out real, quality restaurants, a thing that has actually been happening for some time.
That pretty much sums up my experience. Everything is more expensive and worse at the same time. And it feels like we're being screwed over from every direction and in new ways every day. People keep talking about this tech is supposed to do so much but everything just gets worse.
Private equity bought out the healthcare system where I lived so now it takes months to get an appointment and the quality of care is horrible.
Not that matters since the health insurance I pay hundreds for a month wouldn't cover anything anyways because it's through UHC.
My job keeps getting rid of positions because they want to "do more with less" but it just means I'm doing more work while being paid the same.
If I buy a new car it will spy on me and sell my data. The state DMV already sold my personal info without my permission and now I get constant spam.
Private equity bought out many of the restaurants I liked going to and ruined them.
Third spaces in my city have been mostly destroyed.
The social media sites I used to use were overrun by bots.
I can't upgrade my PC because hardware prices have gone insane.
Gas is like $5 a gallon now so I can't really justify going on any trips anymore.
Private equity bought up most of the rentals where I live and so now I pay much more for rent on top of crazy monthly fees while the building I live in is falling apart because they refuse to fix it. Housing prices have also gone up so much that I have no hope of owning a house.
It's not true that everything is more expensive. Nominal prices are higher, but real income has risen faster than inflation. Purchasing power of the average person in 2026 is higher than in 2016.
Average, or median? Don’t know the answer offhand, but there are a small number of well-known people who are likely to skew the average quite a bit upward, given how much wealth they accumulated during and since covid
Both have increased, but I probably should have said median. I was using average in the informal sense of "average Joe". Median household income adjusted for inflation has gone up 13% in the past decade.
I would just add that consumer products in general aren't the same - practically every business model has incorporated obsoletionism to the point it's practically cancelled out the curve of progress and evolution of tech. We could have nice things.
At least for me, personal agents have been quite useful. If you integrate them well - shopping, filling random government forms, keeping track of my personal life, optimizing training plan and nutrition, planning vacations, etc. I can do more home repairs or renovations myself with help of chat
So basically you get personal digital assistant - very useful - but not changing my life.
What would I need for real transformation though? Food, house cleaning, commute. Even if we had ASI tomorrow, what could it do to change my life, besides taking away my job? Design good house robot to help me with food preparation and chores... Maybe finally self driving cars launched in my area? Or it will take control, "optimize humanity survival chance" and realize we don't need as many people on this planet.
Personal agents are wonderful! You can ask for hiking trails and it will customize one for you, frequently resulting in emergency calls for rescues! Your purchasing/selling AI will arrange sales for you but forget to tell you, leaving you in breach of contracts you didn't know about! Your personal nutrition agent will absorb all the broscience and customize a diet that will result in hair loss, but at least your in ketosis, amiright! And instead of hiring a plumber/electrician/contractor to do work subject to code, you can just ask an AI to tell you how to do it and then wonder why your buyer is suing you for $$$!
(Those are all examples from just the past 4 days, and just the ones that made it into mainstream media.)
This is the core hubris of everyone behind the LLM craze. I do not want to deny their utility there absolutely is some there. But the inherent assumption behind the whole thing is that with a "country of geniuses" problems will melt away. But that's not the case. The reason problems like housing or famines or war aren't being solved is because structurally there's no incentives to solve them not because there's not enough geniuses out there.
And the LLMs exist to further those problems. People can't afford houses? Well, now they won't be able to own computers either. That's a feature, not a side effect. Once the K-shaped economy progresses enough, you won't even be allowed to use the LLMs. They'll exist only to be a tool for the new feudal class who owns you.
You don't need a back room filled with cigar smoke where supervillains coordinate their plans while twirling their mustaches when good ol' class interest does the same job.
Luckily, the AI bubble will destroy a decent chunk of their wealth.
The Gartner Hype Cycle [0] is real, and ubiquitous. We are teetering somewhere near the top of the Peak of Inflated Expectations, about to plunge into the Trough of Disillusionment. A lot of very rich people are about to lose a lot of money, and we can only hope that they don't get bailed out (again).
Then, once the bubble and hype is over, the long, slow work of actually doing useful things with this new tech starts. That's when we'll see it affect everything else.
> We are teetering somewhere near the top of the Peak of Inflated Expectations
I've been hearing this for a year and a half now. That word, "ubiquitous", is getting real mighty challenged. I have tried to insulate myself as best as I could and it just never happens and every other day I wonder if my foil hat needs adjusting. Does yours?
> Luckily, the AI bubble will destroy a decent chunk of their wealth.
Even if all this is true (personally I doubt it), it assumes the bubble won’t burst before they’ve shifted the money from all our 401ks, IRAs and index funds into their own pockets.
yeah, predicting timing on the peak is tricky. Same for all the other tech cycle peaks. If you could reliably predict them then you could reliably make a fortune.
I, personally, think that choosing to move away from hyped tech is probably a mistake. The trick (imho) is to accept that it is hyped and will collapse, and use it while it makes sense to use it. Just don't bet everything on it. Have a backup ready.
> The reason problems like housing or famines or war aren't being solved is because structurally there's no incentives to solve them not because there's not enough geniuses out there.
When I asked why Musk doesn't put his money towards non-profits my Musk supporting friends say it's because he "doesn't believe anything but for-profit entities can solve the world's problems".
So why isn't he solving homelessness, famines, war, etc. with capitalism?
What I like to call at work "this is not a technical problem". There are many problems we could easily solve if we had the right structural incentives. But instead we have politics, processes, layers of management, etc. All that doesn't disappear (yet?) with AI.
I believe that’s actually one of the arguments of the boosters. Utopia will finally be reachable when we get all the human drama and bureaucracy out of the way
I’m not sold, but I’ve heard this line of thinking many times
Human drama is what people use to advocate for themselves. Bureaucracy is what people use to protect themselves. Then you have the side effects but, you may also not like a world without those at all.
As long as the world revolves around the central focus of making as much profit as possible as quickly and easily as possible the same cycle of enshittification will continue.
The question asked by decisionmakers is never "How can we leverage this to make people's lives better?"
It is always "How can we leverage this to make more money?"
And those two aims are almost always in direct conflict.
Companies are not trying to eliminate physical media because it is a benefit to people. They are doing it to drive growth for the sake of growth.
Whatever benefits AI has to offer on a societal level are going to take a long long time to show up, give it some more time. We had smartphones in 2007, and e.g. By way of example, Uber didn't launch ridesharing until 2013, and then that still took years to seep into the public awareness.
By comparison, we're in the early years of AI. We're all still figuring out what this thing even means.
I agree with you, but those examples are going to fall on deaf ears. The pessimistic mindset will say that smartphones were a terrible invention that's ripping society apart, Uber is an unethical company exploiting workers, etc.
I think ridesharing was a net good for ordinary people even if Uber is/was unethical.
However, the counterargument falls entirely flat. "Whatever benefits AI has to offer on a societal level are going to take a long long time to show up". We don't have the luxury of time because AI disrupting communities and livelihoods faster than perhaps any other technology we've ever seen, with the most capital that has ever been put into any tech since the railroads.
Even if you're optimistic we'll eventually get the cancer cures and fusion reactors, doesn't it make sense from the perspective of an ordinary American to regulate the harms like the homework cheating, parasocial relationships, job losses, cyber attacks, etc., especially if it will take years to get the benefits?
Different benefits will show up at different rates. We don't need to wait for cancer cures and fusion reactors. It's already easier than ever to start a small business and use AI for things that would have been prohibitively expensive or time-consuming before. The technology is transforming how scientific research is done, how engineering is done, etc.
I don't see much evidence in the economic data that AI is disrupting livelihoods. Unemployment has been stable around 4% for a while and productivity growth has been at historically normal rates for the past 5 years. I think a more likely explanation is that AI is increasing uncertainty in some industries, which are seeing higher employee turnover, and many of those people are moving laterally into other jobs where their skills are more likely to be useful.
I don't think anyone in this thread has argued against mitigating harms. The OP was purely about current capabilities and the implications for technology in society.
To be clear, I don't disagree with your sentiment that the disruptiveness of AI is causing a lot of grief. But magnitude of disruption is not an indicator of magnitude of benefit (or lack thereof).
Some of the most disruptive periods in history brought about the greatest long term improvements in society. This is the price we pay for changing how things are done, sometimes for better and sometimes for worse. We don't know which one it is for AI yet.
I miss being able to reliably hail a taxi A LOT. Uber may be an improvement in smaller towns where cabs didn’t exist, but in bigger cities it is mostly a downgrade. Rather than walking to the curb and immediately getting a car, I’ve got to dick around with my phone and wait for a car to come to me.
Living on the south side of Chicago, it was almost impossible to hail a cab, and often times if you called for one it just wouldn’t show up. Or I’d try to hail one downtown, and be refused once I told them my destination. I’ve not had that problem once with Uber, good fucking riddance to taxis.
What's frustrating is that at the end of an event like a concert, everyone goes outside to catch an Uber/Lyft, but we then have to match driver to passenger, instead of having a line of waiting cars, each group just gets in the next available vehicle, shows the driver a QR code, instead of the mess we currently have. Why can't we do that? Taxi regulatory capture.
Taxis were terrible and I celebrate their destruction at the hands of the rideshare companies. Even now it’s still cheaper than a cab was (though those early days of VC cheap rides were great).
Uber has been great, I can get around to almost anywhere with a click of a button and have food delivered from a variety of local restaurants at any time of day.
I'm not sure why you think it was a massive mistake? The taxi industry was awful and unsafe, food delivery for many local places was either not available or poorly implemented.
Also many whole food chains didn't have proper delivery for groceries.
Using AI for everything digital, I end up going and doing chores while waiting for the LLM to do its thing. What an insane world, where knowledge work (dreary and uninspired as it is) is automated while I am consigned to scrubbing dishes and waiting for the machine.
Honestly the big reason why I use my dishwasher is that all the dishes would not fit in my drying rack. I'm not very convinced the process of me rinsing and scraping my pans before they go in the dishwasher is any faster than when I lived in apartments when I handwashed everything. Maybe my dishwasher sucks too, but I mean it doesn't hold a candle to the sort of jobs on my pans that require good old fashion baking soda paste and elbow grease. I'm still slaving with this thing. It's picking the low hanging fruit for me in other words.
I feel similar about my robot vacuum. It actually takes a surprising amount of labor facilitating the damn thing. I have to walk through the house and basically vacuum proof it so it doesn't eat various cords, rugs, lifting chairs up and putting them someplace so it can get under, etc. Not sure its really a labor save to when I was just brooming the entire apartment in 20 mins and not having to move anything around due to reach of the broom and it not having a tendency to eat things. Hand vacuum I can be more carefull too about the fringed edges of rugs and don't have to fold them over themselves.
The trick is dialing the things in properly once, and then reaping the benefits over and over again.
e.g., my home is always vacuum-proofed. It automatically runs through all rooms except for the bedroom at night, finishing just before my alarm clock rings, then vacuuming the bedroom an hour after the last person's alarm clock has rung.
This way I never actually notice it, but the home is always perfectly clean as if by magic.
Similarly, if you use separate dishwasher detergent, antisurfactant and salt, and properly dial in your water hardness, it'll be able to clean everything no matter how dirty the dishes are. Including dried-on tomato sauce. Just put the dishes in, and they'll be clean the next day.
Of course it takes a bit of effort to set this up, but then it just works for a long time.
Similarly, I'm a big fan of handwriting software with care & focus on quality, which takes a bit longer at first, but then allows you to trust the software to continue running for years to come.
I've done a lot of vacuum proofing, even getting the magnetic strips, but there's only so much you can do. Like it doing a little wheel around a chair leg is just not as great of a clean as if I moved that chair out of the way. My rugs have fringe so there's no stopping that. Cables I tend to have out have to go away.
It's definitely easier in a minimalist & tidy home. Our rugs have no fringes, cables are either rolled up or permanently installed & hidden, chairs are stored at night upside-down on the tables they belong to.
Sure, it's not perfect, but it's still really convenient.
LLMs could’ve been developed a lot slower without capsizing the entire supply chain and been seen a lot more positively, China is building a lot of good AI with a minuscule of the compute power and it’s fine
there was never a need to rush other than for capitals self interest to control
if frontier labs wanna pace the frontier, they should be honest and return the compute back to the people
China is building good AI by distilling American models. I think that's a great thing, but Chinese progress is currently dependent on American progress, because China lacks the compute to train the largest models (as stated by DeepSeek's founder[1])
i knew about the compute already (he was very public about i believe huawei under delivering the amount of chips he requested), but nowhere in that transcript does he say he's dependent on US progress for his own progress or he's distilling the US to make progress
i don't really think distilling is even that fundamental anymore (nor was the evidence ever that strong - https://msukhareva.substack.com/p/is-kimi-k3-really-distille...), the game is all RL now, and you don't need american progress for that at all (just compute and data, and the data brokers will sell to anyone no problem)
i do think there are black market proxies from china to the US trying to sell "claude" and "deepseek" access mixed together, and i do think that prompt distribution is probably going to be very nice datasets
I'm immensely sad every day seeing all these posts talking about billions and trillions of dollars and where I live (part of the US btw) we don't even have fully reliable electricity and water access. In fact our electricity bills have increased and there is nothing to show for it.
I don't understand why these big labs aren't doing more to improve the electricity situation. Dump a billion dollars worth of tokens into fusion analysis, research, and simulations and try to force a breakthrough. You could buy a lot of good will with unlimited cheap energy that liberates us from fossil fuel dependencies.
Because they increased the demand so much they are seeking out their own supply. Nuclear has consistently been panned for decades because it's more expensive than other power generation.
Mathematics and programming is verifiable and fully controllable in a digital environment.
Fusion research and cancer research requires expensive experimentation in the real physical world. You can’t just throw a billion dollar worth of tokens to handle the problem.
Look at all the roadblocks for data centre buildouts which the industry certainly has a financial incentive to complete as fast as possible.
> Fusion research and cancer research requires expensive experimentation in the real physical world. You can’t just throw a billion dollar worth of tokens to handle the problem.
When the tokens are hooked up to physical devices (robotics of various sorts), then the tokens pay for themselves by being physical workers, or human wage replacement.
And a robocop will follow whatever is passed for law, damn the masses who might vote otherwise. And subjugation will be enforced by token, LLM rule, and machine. They have no union or ethics to complain to.
The complexity of manufacturing is one reason why TSMC is one of the most important companies in the world right now and why Nvidia is the most valuable by marketcap.
If you can make a GPU more efficient and cheaper than Nvidia does you can make a ton of money. Plenty of labs around the world are working hard on this problem and yet Nvidia is still dominating (and the market believes this isn't changing any time soon)
Same things for robotics. It's unlikely you'll have a vertical stack on all the different things needed to manufacture your robot on a massive scale. What happens if some geopolitical spat makes getting rare earth minerals and battery tech from China impossible?
Compare that to tokenmaxxing in programming and math, where the only thing you need is a gpu cluster and electricity.
Electricity infra is not a technical problem. It’s been solved for decades. It’s a political problem, and both sides of the isle have had plenty to do with getting in the way
This is a Western-centric viewpoint. I'm in Asia visiting a few different countries right now, including the big ones, and much has changed with AI and also in general over the past 10 years with cars, trains etc.
> AI is solving mathematics problems (that nobody - including mathematicians - wanted it to solve)
Strange viewpoint. In mathematics any novel result is useful as it expands the scope of other results. Who are these mathematicians who didn't want AI to solve the problems it did? I see trepidation about the future of the field but that is not equivalent to not wanting them to be solved.
I've watched some videos from Chinese commentary on AI usage in the industry,
Some examples are finance people directly asking for API access now so that they can make their own tools. These tools used to be maintained by teams of developers.
Many business making their own customized scripts to make their tasks faster.
From an economic point of view, it seems like there are some productivity increases, but I don't think these can be concluded for a while.
The rational thinking rarely account for second order effects.
Although business leaders rarely care about second order effects, a worse business is not immediately noticeable, compared to hiring cuts and the money it saves
That is not how math works. There are reasons why the Riemann integral is pervasive and the Labague integral is rarely used unless needed. Just because you can do something does not mean that you should.
> Just because you can do something does not mean that you should.
Again, strange viewpoint in the field of mathematics of all places, unlike, say, economics where doing something one shouldn't has disastrous real world consequences. Until it's discovered and proven, one doesn't know what would be valuable and what would not. I bet your sentence would not be said about humans trying to solve problems, no one would tell them they "shouldn't."
Terence Tao and the apparently newly formed Association For Human Mathematics. The post is online-famous now for the line "Mathematicians did not ask for this work to be done."
When I was a young man reading gauss, and thinking before I turned the page, I see a result here, something meaningful! and then I turn the page, and of course it is there. There is no shame in this.
A long time ago, another student of mathematics described it to me as climbing a mountain. No single step is hard, but you must make many steps to get to the point that no one else has made a step.
No need to trust me, just look at how China was 10 or 15 years ago to now. Beijing air pollution, more high speed rail than the rest of the world combined, electrification of vehicles and economy, etc.
Not for Tesla owners in select countries, where their car now drives them around 99% of the time, or the people of Austin Texas, that can now hail a fully driverless Cybercab for cheaper than an Uber!
>Food quality is worse.
Food quality is higher than ever if you want high-quality food!
Is it reasonable to expect AI improvements to filter into all of those aspects of life that quickly? It's only in the past 3-6 months that we've started to see major breakthroughs in mathematics. It might take several years for that research capability to filter out into broader areas like engineering and medicine in a way that impacts the average person's daily life, but it seems highly likely that's coming.
Your pessimistic view doesn't hold up against the data. For a lot of the problems you mentioned, we're objectively in a better place now than almost any other time in history. Of course, things can always be improved and we should never grow complacent or accept injustices. But I would say in a lot of ways there's never been a better time in history to be alive.
It’s amazing how much the goal posts move in every AI conversation. A couple of years ago the average HN thread about AI would be filled to the brim with skepticism about AI ever being able to accomplish anything other than flappy bird clones.
Now it’s comments like this lamenting that the amazing LLMs we’ve had since early 2026 haven’t entirely revolutionized the whole physical world and even global geopolitics yet.
> But hey...on the flip side...a lot of people are rapidly building software (that nobody is using),
I guarantee a lot of software you use today has substantial AI development in recent releases.
Even the Linux kernel is getting a heavy input from AI and Linus is advocating for it. Unless you’ve frozen your entire software stack in 2024, you are using software with AI generated code.
> and AI is solving mathematics problems (that nobody - including mathematicians - wanted it to solve)
Saying nobody wants those problems solved is flat dishonest. The complaints from some mathematicians are that they and their peers didn’t get to put their names on the papers, not complaints that the problems were solved.
With AI software replication, I wonder if shareholder value even increases.
Services accounts for 80% of American industry, what happens when big software companies loses revenue to piracy/recompilation i.e. Adobe, Unity, AutoDesk, tax/law software, etc..
> But hey...on the flip side...a lot of people are rapidly building software (that nobody is using), and AI is solving mathematics problems (that nobody - including mathematicians - wanted it to solve)
That's not fair. To first approximation, most of the progress in human condition of the last, like, forever is implementing solutions that people didn't see value in at the time of creation. Woz built a toy that could barely run software at all. The human genome project was panned as a boondoggle. Penicillin was a mistake.
The first fire pit probably looked like a dangerous stunt to the tribe faced with it, even.
Does that mean everything will come up roses? No. But LLMs only started talking to us for real like three years ago. Let's not just that fast.
I've been thinking about this for 2 years. At some point it will be easier for them to create self sustaining cities - free houses cars public transportation clean food clean water -
Africa is going to look like wakanda in less than 5 years if WW3 gets canceled
>AI is solving mathematics problems (that nobody - including mathematicians - wanted it to solve)
Do people need permission from mathematicians to solve math problems? Is one of the Millennium problems not worth applying a new system to work on it?
I understand the despair people are feeling at AI automating their work, but we have to be serious about the fact that the frontier labs are just responding to natural incentives. Mathematics, coding, all are just applications of logic, and it makes sense that a system trained on inferring patterns from data would work well in that domain.
You are misunderstanding what mathematics research is. Its objective is the development of human understanding of mathematics. This requires open problems to research. AI are “solving” the open problems without creating any human understanding.
At the very least, the single bit of information of whether a conjecture is true or false advances human understanding. It allows researchers to abandon paths which would not have been fruitful.
I'm pretty sure OpenAI's proofs, while difficult to understand, have a net-zero impact on human understanding of mathematics. It took the first Millenium problem years to verify its accuracy. It's only been days since OpenAI released its problems, surely it would take years to evaluate whether the solutions meaningfully advanced human understanding.
It may provide that one bit of information but it could still be a net negative if it destroys current institutions (such as graduate studies) and no new institutions can be developed that provide a meaningful signal of who is doing meaningful work. That was a byproduct of solving open problems that is now lost. These are not my ideas, these are things that Terence Tao and others have written about that were recently on the front page of HN, e.g.:
This is basically the dictionary definition of the word "research". But yes, all the work they are doing is focused on this goal. They write papers for humans to read, give lectures for humans to listen to, have discussions with humans - all with the objective of improving human understanding of the subject.
I don't know, this just seems like gatekeeping of the term "research" simply because you don't like the way it's being done.
Many modern medicines do not have understood mechanisms of action, yet we know they work. Are the people who found them not conducting drug research?
Do we even have a different term for what the AI labs are conducting if it's not mathematical research?
Additionally, I would argue that any work that adds to the total corpus of known facts increases human understanding, even if the path to get to each fact isn't understood. It's hard to argue we'd be better off with less total information.
Honestly it just sounds like you want to complain about things. It is literally in the dictionary definition of the word. We do research so that we understand things better.
What the AI labs are doing is mostly AI research. They are trying to test the limits of AI's ability to do mathematical research. This has been a productive subfield of generative AI research. So far, AI can't do mathematics research unaided - it can't produce both an original result and explain it to people. But they are getting better. Developing tools to aid mathematics research is not itself mathematics research, though it may both require new mathematics research and also contribute to mathematics research.
The difference is the subject area that is being developed. Developing the Lean theorem prover was mostly computer science.
1
: studious inquiry or examination
especially : investigation or experimentation aimed at the discovery and interpretation of facts, revision of accepted theories or laws in the light of new facts, or practical application of such new or revised theories or laws
2
: the collecting of information about a particular subject
3
: careful or diligent search
Nothing here requires human understanding to be the goal. In fact, it explicitly calls out "practical applications" as one of the purposes that can be aimed at.
The whole field of applied mathematics is for the purpose of finding practical applications. Therefore any research conducted in this field is not for the exclusive purpose of human understanding or knowledge.
I'm not complaining, you're just restricting the definition of mathematics research in order to discredit what the AI labs have achieved.
OpenAI to this day is not aware if they solved or not Navier Stokes.
Mathematicians are not gate keepers but I understand the frustration. If someone claims they spent 30 million dollars to solve one of the most difficult problems yet there’s no new knowledge, understanding, nor certainty. Then what’s the point besides marketing?
I think one of the hopes is they'll get smarter than us and able to solve real world problems better - disease, war, poverty etc. But they are not really there yet. Give it a little while.
What is up with this attitude in this thread? AI has taken off programming basically this year, the end of last at best. The first truly capable LLM, ChatGPT, isn't even 4 years old yet.
Just a crazy amount of doom and gloom all around this thread.
In other words, it's taken a highly enjoyable and lucrative vocation from vast swathes of workers - a lot of which make up this sites audience! - and instead concentrated more productive power in the hands of a few megacorps.
It has made it far more accessible to a lot more people. Being “highly enjoyable” and lucrative” is why people try to gatekeep it.
The fact that the means of this capability is in a few hands does suck. But I’m increasingly optimistic that it’s going to be commoditized through open weight models and performance improvements. There is enormous demand for innovations on these fronts.
I'm a rare individual that has consistently made choices to earn less money in order to have more enjoyment. I would clock fewer hours than they wanted me to during college for my internships, I left a FAANG to go back to startups, I quit a startup to work on volunteer work and try to bootstrap a business.
At multiple jobs I've noted that I felt the pay should have been less. Not so that the employers could take a bigger slice of the pie (in fact, my preference is to be in an anarchist co-op, but I understand that's a wild take and might be terrible in practice). My hope was that we could filter for people that genuinely want to do the work and take on fewer people that were there for the money. At the same time more money is more better. And I like having money. It's complicated.
As I've said many times before, I didn't come to Silicon Valley to get paid. I came here to make computers do cool things alongside other people. I need income to do that, but I try to make sure I'm in a situation where the income isn't something I need to think about all of the time. If you're doing mercenary work that's not possible.
I think being able to accomplish more from places that are a fraction the cost of Silicon Valley will also help achieve the goal of doing things for reasons other than the money.
I think the amount of protectionism we’ll see could be absolutely enormous. Not only do incumbent corporations not want to lose share, incumbent countries won’t want to share either.
It's kind of weird to say it's "taken off" recently. Yes what we have today is far better than half a year ago and solving new real use cases, but that was far better than we had half a year before that, and solving new and real uses, etc, for years now. On their best usecases LLMs were already enormously useful years ago, and it's still very spiky.
As a user of software I see no improvements. What I see is that I need to download software more often, it brings more features I don't need and it has been as buggy as ever (if not more). And I can say that about Operating System I use, social networks, games, etc.
Exactly. Where are the results? If AI is really producing value at 20X the rate software engineers produced it earlier, we should be seeing 20X the number of software projects released now. Or they should be getting made 20X faster. Or they should be 20X better. Or the projects should have 20X fewer bugs. But none of this seems to be happening. The only thing that seems to be happening is we're making the same crappy shit, but with 20X fewer people.
I don't think it's surprising that it hasn't improved life much, AI is new and change takes time, but the fact of the matter is that it also made things worse in some ways, ways that I think were entirely avoidable, if only the incentives weren't so warped.
The fundamental malignancy here is that powerful people have become disconnected from the foundation of society. They have suffered from model collapse. They have been talking to each other in circles for too long, and doing things that make no sense to anyone outside the elite group, and because the systems that concentrate power to the top are so strong now, they continue gaining power even while messing everything up in a nominally free competitive market.
People need to eat, sleep, be healthy, have shelter, feel loved, be in communities...
Coding has been transformed but none of those areas of our lives have been substantive transformed by AI yet. Whoever does, will create a lot of value for the world.
I think that's a bit disingenuous. We've only begun to apply LLMs in the real world and a lot of industries don't move very quickly. Not to mention that we're still very hardware constrained, we don't yet have the equivalent of an ESP32 chip we can just stick into any piece of electronics to provide state of the art LLM responses. But we'll get there, soon.
The funny thing about comments like this is that the moment LLMs are going to replace everyone in the warehouse or the trucking industry, we suddenly have to do something against the LLM. But clearly they can't actually do anything meaningful and have been a waste of energy?
So just let them spend their money and they will eventually fail and bankrupt, right?
This is the paradox. For software engineers, the current crop of “AI” tools is magic. A huge accelerator. And then they assume the rest of the world will get the same boost. However this is not the case at all. Claude code is great because it is trained on extremely well documented git histories and stack overflow. And outcomes are binary: the code works or it doesn’t; the PR is accepted or it isn’t. Not much else in the business world is like this at all. No training data with this kind of sequential development information. And no real black and white outcomes as such.
I would love it if we can just take the win on having created an amazing tool set for coders. Can’t this be enough?
"Fuel is more expensive." "Wealth inequality is accelerating."
If Americans cared about that they wouldn't have put the far right into power. I think its a little naive to expect AI to address any of those things when they are not in the realm of AI to address. These are political policy and Americans simply vote pro-capital owning class and anti-working class for misguided reasons.
In the workplace, AI can be quite a time saver so I think people getting no benefit is being disingenuous.
It's actually sad how much whinging doomers are doing and how you're all agreeing each other like some crazy death cult. I think you all want the world to got to hell just so you don't have to accept change.
> "...how much AI has either not improved life at all or actively made it worse from the perspective of the average person"
This is empirically, obviously and blatantly FALSE.
Claude and ChatGPT alone have 1.5 billion active monthly users. Do you REALLY think all those people are using it because it all sucks and is useless? That's 1 out of 8 people on the planet and it doesn't even count the million other AI companies out there (Google and Meta included).
Off the top of my head:
- Have you seen the TikTok videos from normal people talking about how Meta's Muse amazingly helped them do stuff like sell everything in their garage online with just a picture, or proactively helped them not miss appointments? I have.
- Have you talked to people who've used Claude to manage their small business or consulting by dealing with emails and rescheduling tasks? That's what my therapist friend is doing - all without any sort of technical help either, which is incredible.
- Have you been personally astounded at ChatGPT's ability to diagnose the problems with your car simply by your rambling description of a noise it's making? I have. It's really nice to go to a mechanic and have a general idea of what might be wrong.
- Have you talked to non-techies who solved their computer problem with the help of an AI chat giving them the right prompts to enter? It's so adorable how proud they are! Unlike getting help from people, AI is willing to walk them through their issues step by step, answer their dumb questions and not make them feel like they're stupid just because they're not technical. It's honestly heartening to see someone feel like they understand how their computer works for the first time in their life. (Actually, reading this, I realize it's an analog to me understanding my car for the first time. Heh.)
This stuff isn't even a smallest, tiniest, microscopic sliver of what AI is doing for people right now.
You're just being willfully ignorant or obtusely closed minded because you don't like change. Too bad, things are changing as they always do. Accept it.
> Imagine what a "country of geniuses in a datacenter" will be able to do? Apparently...nothing.
I suspect the real world change is going to happen unbelievably rapidly and isn't far off. Right we don't quite have the hardware to embody AI in a generally useful way, but that's coming fast.
Within a couple of years arguably any task that requires just knowledge or intelligence is something you'd be better off giving to an AI than a human. And decent humanoid robots are maybe a decade behind that.
In the same way people growing up in the mid 1900s saw rapid changes in their way of life with planes and cars replaces boats and horses, we'll see something similar with robots replacing human labours. Right now we're still in the phase where people don't think humans will ever be able to control flying machines and that cars are reliable and slow. Things will change.
I think people in the LLM space are being far too impatient with the general public and their adoption. And I think they need to start thinking harder about building products that genuinely enrich people's lives. Look at how long home computing took to establish itself with the public, then the internet, and finally social media. None of these trends happened overnight, and some were accelerated by other technological advances (social media-> smartphones).
The adoption of these technologies spurned ton of innovation in these spaces, and I feel like with LLMs its the first time I've seen genuine frustration from the industry directed towards users for not adopting the technology fast enough. This attitude did not exist in the 1980s, it didn't exist in the 90s, and it didn't exist in the 00s/early 10s either.
Do you blame them for getting impatient? They’re burning cash and quickly running to a point of never being able to justify it. The hardware they buy will be replaced with even more expensive hardware before they start to make a profit. It’s not sustainable and the market hasn’t figured that out yet. At least they’re starting to question it.
Whether you like it or not - technology causes compression of life cycles. You have a hit product? Great - I expect you to have crazy high growth rate. That’s the expectation of a world that is so deeply interconnected.
Consumer-ready LLMs have been widely available to the public now nearly 4 years if you start counting at GPT 3.5. There's certainly some compression here but Karpathy does a good job of outlining how silo'd those experiences are: the bulk of the product penetration (based on headcount rather than token usage) is "derpy chatbots" (his words, not mine). Most of the actual token spend is likely done on the STEM side. We're talking about a technology today that currently only has significant penetration with <25M users.
People need to be patient and put in the work to build great products, rather than lament its slower-than-desired (albeit still quite fast) adoption curve.
Yes. The AI industry has to deliver some substantial value to the average citizen soon or there will be political trouble. Food, clothing, shelter, and medical care have not improved yet. Nor does there seem to be much AI can do in the next few years that will help with the basics.
We can have unlimited new entertainment content for passive consumption. That's about it so far. 17,000 new Asian vertical dramas a day now.
I was listening to a Verge podcast in which Nilay Patel remarked that of the AI companies started pushing for universal healthcare, AI would become a lot more popular. But no signs of this.
Please don't be snarky or curmudgeonly in HN comments. You can make your substantive points without either of those things, both of which the site guidelines ask you not to do.
You truly have become an insufferable human being. You actively censor the handful of good comments on this site, probably just because it makes your job easier.
The comment you're replying to is one of the handful that is actually useful, and the fact it was at the top of the thread until you hammered it down is proof everybody else thinks so too. But I guess we can't disturb the techochamber.
Seriously, just fuck you. There is a reason more and more people hate you as time goes by.
Not sure if that's a fair criticism dang? I sincerely wasn't trying to be snarky or curmudgeonly, but I did want to make the point clearly that none of the benefits of AI have materialised in the eyes of the average person, only the downstream costs. To my eyes, that's a salient, reasonable, and wholly representative position to share.
Did the writing also contain emotion? Yes. As all good persuasive writing should.
As the other moderator, I'll give my own perspective on this. I haven't spoken to Dan about it but I completely understand why he thinks this is not the kind of comment we want to see on HN.
The main issue is that it doesn't engage with the substance of the post, and in fact, it exemplifies the very point the post is trying to make: there is a big and growing gap between those who recognize the benefits or potential of AI and those who don't (or who only see negative effects or negative trends in society).
I'll make it clear that I don't think your complaints or concerns are unfounded or invalid. There's merit and substance to all these issues, and it's fine to discuss them on HN. They are discussed all the time here, and they can and should continue to be discussed in the future.
But an important part of being the kind of curious commenter that we want everyone to aspire to be is that we can look at topics holistically. We need to be able to look at the opportunities and risks together and discuss ways that the benefits can be captured and shared throughout society, and the risks and costs adequately recognized and curtailed. Most people I know of in the tech industry are taking these matters seriously.
It's a vast and complex topic to discuss, and here's not the place for me to get further into it. But the point is that it's a bigger and more complex topic than your comment adequately recognizes, and we need people to always be making the effort to be thoughtful and curious rather than fulminatory or denunciatory in their participation here.
(Also: I can understand you may not easily recognize that your comment breaks the guidelines; it's common for people to not recognize that their words seem more inflammatory when read by others than they intended them to be.)
So, if I say, there is a big and growing gap between those who recognize the benefits or potential of my poop machine and those who don't, and you say, wtf is this shit everywhere, I can just say, you doesn't engage with the substance of my post, and in fact exemplifies the very point my post is trying to make
You might have a point if the original comment focused on the negative consequences of AI. But it doesn't; it's just an enumeration of grievances about negative trends in the world, most of which predate the emergence of AI, and/or are orthogonal to it.
The people who are optimistic about AI believe it can help to address or reverse these trends. It's fine to debate the merits of that belief, but the original comment wasn't trying to do anything like that.
Many, many people here have already been sucked into the TESCREAL orbit, and have a quasi-religious belief that AI will be the most important thing to ever happen. Anything that challenges that stance is anathema.
It honestly escapes me why you would judge the juxtaposition between everyday life of everyday people and a very specialised minority as snarky or curmudgeonly. Was there too much emotional impact?
There are so many comments on HN that would justify your attention with regard to this specific rule. This is not one of them.
The fundamental question I have that honestly I haven't been able to find any answers for: With all the talk of LLM-based AI capabilities "going vertical", what evidence is there (for or against) that LLM-based approaches won't eventually "hit a wall", that is find some aspect of intelligence where humans will still have primacy, and no amount of scaling will change that.
E.g. some prominent folks definitely think LLM-based approaches will hit a wall, perhaps LeCun most notably. I also read a guest post on Terry Tao's site (which I really liked) that argues that, for all the very impressive recent AI results, they still operate within the "convex hull" of their training data: https://terrytao.wordpress.com/2026/09/13/happy-those-able-t...
I'm just curious if there is any actual data or evidence that takeoff (i.e. RSI, "the singularity", whatever you want to call it) is inevitable with current approaches.
But when you squint your eyes, LLMs are just a least-effort manifestation of the wider neural net direction we’ve been chipping away at for decades. I don’t give a fuck whether LLMs are here to stay; they’re proof that scaling neural nets gets you pretty far.
The real question is: will neural nets hit a wall? As a human, I certainly hope so. But I’d bet that they won’t. I’d bet that we’re igniting the ~~atmosphere~~ universe in slow motion.
I don't think there is evidence in either direction (takeoff impossible vs inevitable).
But the "convex hull" idea always came across as wishful thinking and hand-wavy pseudo-science to me.
Here's a rebuttal [0] of it from Tim Gowers (uses another math metaphor, but still clear and the whole thread is worth a read):
>...the popularity of the idea that AI "merely explores the convex hull" mystifies me. Why wouldn't it be more like the subgroup generated by existing mathematical knowledge? (That's also a metaphor -- I'm not saying there's a group structure.)
>The convex hull metaphor suggests that AI isn't having radically new ideas and will eventually run out of things to do, whereas the group metaphor suggests the possibility of building up in small steps to something well beyond what you started with, with no obvious limit.
"running cyber attacks and defenses at machine speeds"; worth noting that LLMs (even the vaunted frontier models) are exponentially slower at cyberattacks or defenses than your average WannaCry or Splunk automation from decades ago. Its this sort of delusion world the AI bros live in that is part of the reason there is a disconnect between what they think should be happening and what actually is.
933 comments
[ 4.2 ms ] story [ 244 ms ] threadThere is in fact no indication of this, not evem the precious benchmaxxed benchmarks ya'll love to reference.
There is however a exponential curve of slop, and an ever increasing number of peoples who's minds are completely captured by these things.
I'm not arguing that they cant write code, or write a proof. Its just not written or designed well and is absolutely brutal and soul crushing to work with. Look at these proofs they're producing also, they're millions of lines of Lean that are impossible to reason about.
The way we're using the term 'verticle' to describe a curve means we're not being honest about this. This curve can actually be plotted, you can go look at the curve. It is not in fact 'verticle'. Each model release is climbing single digits on benchmarks it was overfit for.
In the last 9 months or so llms have gone from just another useful proof tactic (like grind or sledgehammer or sat solvers) to being so good at writing proofs that I don't even bother to try myself anymore.
Caveat: that same tiny group is employed by the AI vendors, meaning it's in their financial best interest to make it sound like the curve is going vertical.
Do you still write code by hand?
Considering the investment in AI, the lack of moat, and the increased inability of any of the big players to come even close to profitability (with OpenAI already breaking the "ads" emergency glass option)... perhaps he meant a tiny group is already seeing the line crater
AI models can dismantle billion dollar industries. They can reverse engineer Adobe and Microsoft products that once were their moats and titans of their industry.
Why do you think they'd need to lie?
LLMs are great and will get better but we have not seen an open source office suite built by LLMs come online yet, and even if there was one I'm pretty sure few would adopt it, cause that's not the only moat, LibreOffice has been around for decades.
Let's say, for the sake of argument, that the models are some multiplicative factor better on the inside.
Doesn't that mean the demos should work?
None of it does anything particularly amazing, but that's kind of the point -- the programs that people are generating are for the use cases that are too personal, trivial or specific for anyone to have bothered making a software package for.
Amid all the discussion of sigmoid curves, and where the "LLM wall" will materialise, I think few people would have predicted that the real wall in LLMs would end up being humans' capacity to verify the output.
What I fear is that people simply eschew human review altogether, considering we're talking about the industry that came up with the "move fast and break things" credo. Human review of LLM-produced code where I work is already a farce, and we're not special enough to be one of Karpathy's 5,000. I do my best to manually review anything that's my responsibility, but I'm literally one of very few people left working on my team, so in practice what happens is I submit PRs that are at best glossed over by completely unrelated teams for security, malware/prompt injection, and other serious concerns. Quality insofar as vetting others' code has completely gone out the window and it shows in the number of bug reports that come back, often themselves written in Claudease. This is all on top of everyone cynically phoning it in in the first place, due to the omnipresent sword of Damocles that is additional AI-driven layoffs.
Worse yet all the incentives point to this being the most economically viable thing individual companies can do. I think it goes without saying some type of regulation here is urgently needed, and that an unexpected cause of an AI bubble pop may end up being that humans simply aren't able to keep up with the pace of the output - leading either to precautionary plateauing of capability, or major liability risks related to a decline in quality.
For what it’s worth, any time I see that feature, it’s all farts.
There's no reason not to trust it because since they trust it, they don't percieve anything wrong. They just take it and whatever happens happens, no annoying conflicting info and certainly none of that old grampa thinking and working and giving a shit about anything.
But nothing else matched that. I was looking up all kinds of resources and that hadn't been the conclusion from any of them. It turned out Google AI was just wrong and put us in a bad spot for even looking at those results.
If you then follow up with the thing using the little chat box, it actually kicks off a non-trivial model result, which can sometimes completely reverse the response. After all, your search results are usually based on a few words.
It very, very routinely points at articles in its RAG "citations" that simply don't support the (modestly likely to be correct) assertions.
It routinely recommends Mac software to iOS users until you ask it to confirm there's an iOS version.
It still claims widely used software has functionality that is non-existent, based on roadmap/wishlist/feature request/unreleased fork type pages.
It is better than it was but it is not adequate yet.
I think LLMs are valuable and spend most of my professional life working with them.
I do wonder thought whether there’s a Ponzi scheme aspect here where as long as the “frontier” can keep outrunning human review and comprehension, LLMs are always going to looked way more valuable than they are and the bubble will continue.
This started with deep learning, expectations weren’t met and people started looking for value, then GPT came out and people got wooed again and forgot, then coding, then math, cyber, etc. As long as the dust doesn’t settle we never have to reflect on all the shortcomings and can just stare mesmerized at demos.
Their input is just working on the "plan" itself, the spec or the request, iteratively, and so on, while the AI works fulltime on automating the delivery of that thing, improving the systems that are delivering that thing, asking questions about what exactly is being built and how it should work, and so on.
A lot of the difference in how people experience AI is ambition and scope and how much you just tell the AI to 'automate your factory factory' in Factorio terms vs 'hand mine the coal and hand-craft the science packs' in Factorio terms.
EG, if you don't know how you'd speed up the process of building a web browser from scratch (compared to the current way you've been working with agents), you would ask your agent to help you figure it out and design that process itself, rather than just try to do what you're currently doing on a larger scale. You'd literally prompt "how can we make it realistic to build a whole browser when I've been just sitting here planning small changes with you" and then let it act on its response.
> Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt.
No, they can't, at least not any semblance of quality. The cases we're seeing where this does kinda work is in ports and translation where all the rules are already documented in the best specification language possible with a way for the LLM to verify itself: code. We saw this close to a year ago now with Cloudflare and NextJS
> The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about
These are operating on different capabilities - AI's ability to answer informational queries as a chatbot frankly sucks and can't be trusted without verifying it. I run up against this every day. A problem with a verifiable answer on the other hand it's very good at solving. He knows this (his next paragraph) but he's putting them on the same scale of "stressing the system" to attempt to add proof to his introductory claim
And then there's the completely unverifiable scare that there are internal frontier models way beyond anything we've seen "swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics"
I dunno. I haven't been able to set up OpenAIs remote codex connection, their shit is buggy as hell and the web UI keeps crashing and making messages disappear. Is this what their internal superhuman "Things that would have taken top professionals in the industry years of work" looks like? Granted Claude has been really smooth, but still..
Defined testability.
> swarms of thousands of agents collaborating over weeks on software mega projects
If you have defined testability, running a single agent, in a loop, over a goal will achieve the same results.
> minting zero days, running cyber attacks ... Granted Claude has been really smooth, but still
The old adage about monkeys writing Shakespeare applies here. Agents are just better monkeys and if the goal has a well defined outcome it's fairly easy to "Brute force" the way to an answer.
There’s no reason to think this.
There are lots of good reasons to think this.
The first is that LLMs used to be bad at each of these nerd things, then toppled them like dominos. There's a pattern over time.
Another reason to think this is g. Across every known measurement, human (and animal) intelligence is convergent. Being better at one thing correlates with being better at another thing. Although the reason is not perfectly clear, the pattern is well established, and seems to apply to LLMs too - GPT-6 is smarter than GPT-3 at everything, not just math. There's no reason to expect different for GPT-9.
LOL. Human (and animal!) intelligence is notoriously jagged. We're all basically idiots except for very narrow areas where we focus.
If anything, in some domains frontier models have become worse - claudisms and chatgpt idioms are making them notoriously bad at prose without a considerable amount of prompting and tweaking.
However there are many other domains which have verifiable rewards in the process of learning them, despite their overall impact not being verifiable. For example, the life sciences, an LLM could be given access to data about an organism, and then make predictions about how a drug or gene therapy will affect that organism. In economics, LLM's could create models of behavior, evaluate predictions over time and see how well those predictions match reality.
>claudisms and chatgpt idioms are making them notoriously bad at prose without a considerable amount of prompting and tweaking.
I think this is because a lot of people genuinely like the claudisms, even though a small minority of technical people don't.
Anecdotally, creative writing communities do report drastic differences in quality between models (it is my understanding that Gemma 4 is especially praised) so there is something to it being verifiable but it's a far cry from being something that can be distilled into objective math/code-like evals. One could even say it comes down to vibes.
>For example, the life sciences, an LLM could be given access to data about an organism, and then make predictions about how a drug or gene therapy will affect that organism.
Yes, biotech companies are doing this right now but ultimately they are just predictions and you still need cold hard biological data to ground them and iterate on. That's slow and expensive, especially for the juicy fields (human biology, food and crops, clinical trials) and you can't just plug billions of VC capital into the pipeline and hope for RSI. Physical (gotta procure all those labs and their equipment), human (gotta hire and pay specialists to run experiments), biological (gotta wait for organisms to reproduce, drugs to take effect, crops to grow), regulatory (gotta convince agencies that your fancy new drugs are legit) bottlenecks get in the way.
And biology is one of the easier fields where a path to RSI (if not accelerating) is conceivable. Good luck iterating on macroeconomics data where there are no replicates and 'experiments' take literal years if not decades.
>I think this is because a lot of people genuinely like the claudisms, even though a small minority of technical people don't.
My uncharitable take is that the abtruse jargon makes people feel smart for understanding it and gives a sense of belonging (as a closed circle of initiates who understand LLM cant), just like rationalists love to repackage old or unsavoury ideas under new nerdy smart sounding names.
https://vulsar.ai/benchmarks/creative-writing-v1/#cohort-ove...
>you still need cold hard biological data to ground them and iterate on... you can't just plug billions of VC capital into the pipeline and hope for RSI
Capital can get you a lot: access to wetlabs, historical medical data, ability to run simulations, creation of autonomous bio labs which run thousands of experiments in parallel. An LLM, in training on this flywheel of data can take in more information than a human biologist would be able to read in multiple lifetimes.
>Good luck iterating on macroeconomics data where there are no replicates and 'experiments' take literal years if not decades.
The better an LLM gets at simulating human behavior, the more accurately you can simulate economic policies and their effects via running simulations of agents in virtual societies. There are many routes that models can take to create a broader sense of the world without needing the world to provide data.
I would really take those benchmarks with a pinch of salt. We already know benchmaxxing is a thing for evals that are objectively verifiable like code, and coding models that seem to perform well on benchmarks don't pass the vibes test, imagine all the caveats that come with such a subjective test like creative writing, not to mention control issues (are they the same human writers across benchmarks? Are they the same testers?)
>Capital can get you a lot: access to wetlabs, historical medical data, ability to run simulations, creation of autonomous bio labs which run thousands of experiments in parallel. An LLM, in training on this flywheel of data can take in more information than a human biologist would be able to read in multiple lifetimes.
I don't disageee and am in fact excited to see the advances, what I am arguing is that it is going to be slow because the hypothetic feedback loop that would improve the model is bottlenecked. There are actual physical and biological limits in our universe. Unless VCs find a way to make mice embryos and crops grow 10x faster (among the many, many other things that would be required for acceleration) the trickle of data is not going to be anything like the accelerating returns imagined for math and coding. And that's assuming biological RSI is possible which has yet to be proven - current foundation models are interesting but they are still specialized and there is no clear, obviously uniform way to encode the many, many modalities that exist in biological data.
>The better an LLM gets at simulating human behavior, the more accurately you can simulate economic policies and their effects via running simulations of agents in virtual societies. There are many routes that models can take to create a broader sense of the world without needing the world to provide data.
You wouldn't need to 'just' simulate human behavior (good luck with that btw) but the entire political and economic conditions in which they live. Even assuming this is possible the amount of compute you'd need to run would be astronomical. Given how all RAM capacity for '27 and '28 is near sold out, hyperscalers are running out of cash flow to finance buildout and Sam Altman complains when users say 'please' to LLMs I don't think this is in the near horizon.
And that's a massive intelligence gap!
It may just be that writing is mostly saturated, and the remaining perceived gap is not quality but some kind of "cultural fit".
AI writing sucks because there's no intent or choice behind it. It's human thought, blended by linear algebra into pink slime and extruded into sentences.
This is inherently self-contradicting.
Even if we do accept that models have some “g” and that it correlates with all domains, there’s no reason to think that the current trajectory of LLM intelligence will grow more quickly or even as quickly. In every other technology we have an early period of slow growth during experimentation, and then a rapid period of growth, and then stagnation.
For all we know, 2026 could be the year of stagnation. It’s possible that we will reach finishing returns. If thats true, then it won’t matter how much that task correlates with overall g, because g won’t be increasing that much.
It's worse than that. I am one of those nerds (a programmer), and I see that LLMs are shit at the things I care about. I have zero reason to believe that they will be good at other things either.
I don't think this explanation is sufficient:
For example, many nerds care about 3D printing. On the other hand, my experiments (and the experiments of many nerds who I know) to let LLMs create files for 3D printing lead to horrendously lacking results.
Or many nerds care about linguistics or puns. Whenever a new LLM comes out to which I have easy access, I do the test, and let it explain some specific German jokes to me that are based on convoluted puns in the German grammar. Until now, no LLM that I had access to could give satisfying explanations of the puns on which these jokes are based.
Or even for coding (many nerds do care about elegant, sophisticated code): the LLMs that I could test were already overchallenged with the following task: I had written some code that is a very "artisanal", "clever" improvement of an algorithm over the version that one would find in a textbook. The task for the LLM was simply to write some code comments/documentation about the mathematical ideas upon which my improvement over the textbook version of the algorithm is based. It wasn't capable to do this. On the other hand, for a junior programmer, I would in such a situation expect that he goes through every single line and thinks through the algorithmic ideas so that he can learn from them.
---
Thus: even for "nerdy" topics, it is very easy to find tasks where LLMs still suck. So, my hypothesis about the nerds that you mention in your post is that these nerds are rather people who want to believe (with religious fervor) that the current LLMs are exceptionally good instead of just looking into a slightly different direction than where the tech billionaires want the users of LLMs to look at.
Hippies live at the center of technology and humanities.
Nerds sit purely in technology.
Try again; astra has been driving fusion for me. I took a screenshot of a self watering cat grass bin on makerworld — not even the stl, a screen cap of one of the photos attached to the design — and it made me a parameterized version in one shot.
Its initial design was less than ideal for printability and it reworked the design correctly when asked to consider what it originally produced through the lens of printability.
Autodesk bundled an mcp server with fusion in the last few weeks and i expect things to get even better going forward.
I have no access to Astra, but with GPT 6.1 Sol, the results with respect to attempting to create STL, OBJ or even OpenSCAD files by using image references and a prompt were really bad.
The most passionate, intelligent people are usually a decent indicator of where culture is going to go.
Many very smart people who I know are quite skeptical of LLMs and the hype around them. In my observation/echo chamber, the people who are very into LLMs are rather slick, career-minded people who love to present themselves as trendsetters for the "next fancy thing".
So you’re wrong pal, sorry.
>The most passionate, intelligent people are usually a decent indicator of where culture is going to go.
That's a naive view of cultural determinism. The software landscape would have been very different if a handful of lawsuits had gone one way or another. Hell, even Unix and Linux were hobby projects that accidentally got big.
Also crypto is a grift lol, one of those is definitely not like the others
Unlocking mastery of those domains would lead to incredible gains in all others. Whereas a fundamental advancement in many other domains doesn't spread as readily.
That’s good for corporations who need custom software. Every large business has a massive backlog. In this area reducing the cost will massively increase demand.
But where’s the broader value?
Consumers and business already have too much choice in off the shelf software.
Ultimately most people just need word, excel and maybe 4 computer games a year.
It needs to a potential field with practical, economical, "real life" attraction well. I.e, robotics, real economic efficiency gains, manufacturing novelties.
The worry is that the 1% is attracted towards a non-practical money hole. I.e: Token burn for the lols, sophisticated software systems that dont provide actual value outside of giving NVIDIA cash.
Is it effective? I'm not sure, cause if it was I don't see why OpenAI & co are not releasing such a project instead of demos.
We have already reached "peak LLM" in terms of what normal people realistically need it to know or reason about. In fact, I'd say we reached that point about 1.5-2 years ago. There are two other barriers that remain unsolved:
1. They're less dependable than humans and can't be meaningfully punished or forced to make up for mistakes, so you can't really replace humans with them, not without having a human babysit.
2. Most people don't really have a special need for an LLM in their life. They may like that it answers questions or helps you polish a resume or, I guess the labs' favorite, help you make restaurant reservations. But it's hardly a necessity.
It'd be kinda funny if we create superhuman AGI and then no one has any real use for it, perhaps except for military murder-bots. There's always market for that.
Even ones that cannot be easily dismissed (like the agent at the top of Google search) are often subtly wrong in a way that requires significantly more effort on my end to parse the loquacious output to determine where the inconsistencies are.
Look, can they be useful to generate the html or jscript for humanity’s 9 billionth iteration of a web form? Sure. But man, I wish sanity had reigned and people had applied ML models in general to more substantive and beneficial projects.
(Before anyone chimes in, I’m familiar with implementations of ML applied to esoteric domains; but by their very nature these don’t get all the news cycles, or hiring, or any of the other absolute insanity that the domain seems to contain)
Paul Krugman (1998)
• "The growth of the Internet will slow drastically, as the flaw in 'Metcalfe's law'—which states that the number of potential connections in a network is proportional to the square of the number of participants—becomes apparent: most people have nothing to say to each other! By 2005 or so, it will become clear that the Internet's impact on the economy has been no greater than the fax machine's."
Clifford Stoll (1995)
• "We'll soon buy books and newspapers straight over the Internet. Uh, sure."
• "The truth is no online database will replace your daily newspaper, no CD-ROM can take the place of a competent teacher and no computer network will change the way government works."
• "Lacking editors, reviewers or critics, the Internet has become a wasteland of unfiltered data."
There's a massive tendency amongst even intelligent people to completely dismiss the potential of new technology based on that technology's current observed capability.
That clearly didn't happen :)
> In the next decade, they will do assembly-line work and maybe even become companions. And in the decades after that, they will do almost everything, including making new scientific discoveries that will expand our concept of “everything.”
https://moores.samaltman.com/
Your car is still the same. Your dishwasher is still the same. The train you take to work is still the same and never comes on time. Fuel is more expensive. The roads are still congested with traffic. Your kids are (probably) doing worse at school. Food costs more. Houses cost more. Rent is higher. Buying a computer or a PS5 is more expensive. Politics is still full of mentally unstable people. The environment is getting worse. You will still die of heart disease or cancer. Food quality is worse. Your kids can't find work. Wealth inequality is accelerating.
But hey...on the flip side...a lot of people are rapidly building software (that nobody is using), and AI is solving mathematics problems (that nobody - including mathematicians - wanted it to solve)
Imagine what a "country of geniuses in a datacenter" will be able to do!? Apparently...nothing?
I'm noticing this even in industry. Yea, we can build and iterate on ideas 100x faster but has this turned into anything meaningful? Not really... Hell, look at the big companies spending billions on inference... nothing shipped and things are still buggy as ever. Where is everything?
I find 2026 much more repulsive compared to 2016.
Not only from the big companies. Where is the explosion of disruption ?
If I look in the "Library" at images I have generated, scroll down a bit and click the image, all the images turn black/empty, that image then shifts to the top of the page.
Putting it at the top is fine, but it should be done _after_ one closes out of the library.
* Higher load on senior engineers who are about to burn out hard from fighting an endless battle against unreviewed slop people keep trying to yeet into production
* Less junior hiring and suppressed wages
* Culture of uncertainty and fear
* Computers are really expensive now
* Some people now live with diesel generators and gas turbines making their neighborhood hazardous to their health
Gosh, this has been my entire experience with AI, and I’ve never seen it this well articulated. Thank you!
I think some of the most impressive stuff to show someone from 2016 might be LLMs doing a summary of complex topics, followed by self-driving (actually) cars and intricate robots... but other than that there is no "future magic" touching the lives of every day people going on.
Actually I'm pretty sure if you showed someone the latest and greatest from these big AI companies, these AI companions that can book flights, people would go "Oh, so its like a smarter Siri" and proceed to not care at all lol.
I have no doubt startups in particular are able to get away with using LLMs to produce sloppy code and ship. But this trend was on-going and sped up due to the infiltration of LLMs and marketing.
You do realise markets are highly imperfect in reality? That nonsense in econ books is fantasy.
e.g. Why do Rust and CPython still have over 5,000 open issues each? At some point, will the superintelligence fix all the bugs autonomously? When is that going to happen?
https://github.com/rust-lang/rust/issues
https://github.com/python/cpython/issues
What about new features? In every thread about Rust, people are clamoring for new features and libraries. People debate why there can't be a batteries-included standard library like Go.
So now that LLMs are so smart, is that a solved problem? Or are we still bottlenecked on the expertise of a few experts?
---
Also, what happened to AI's passing the Turing test? I see tons of slop articles that are obvious to most HN readers. They haven't fooled us apparently -- the longer the interaction, the more obvious it becomes whether it's an AI.
(That said, I use AI all the time and find it to be a great tool. IMO is is a tool, not an autonomous intelligence with a will of its own.)
More noise. The world is just so noisy now.
Has any technology made an impact on physical world faster?
You are allowed to criticize it, just not because "the progress has been too slow". It was not.
Progress between Copilot and something like a gpt-6.1-sol in a good harness is at least as big as few Android versions.
It reminds me of the Ezra Klein quote that “OpenAl would need permits to cover its parking lot in solar panels, but it can accelerate into recursive self-improvement, as best I can tell, whenever it so chooses."
Likewise when you see other such shortcomings in our world, ask what this incentivizes for otherwise. Then things will become obvious why they are how they are, from housing, to transit, to feeding people, to healthcare.
The priority is profit, not people.
Things worked better before we started putting bad software in everything. As a software engineer, I know what good software looks like. The fact that almost all software is now bad -- including the software that drives AI and the software that AI generates -- makes me want to flip the table. We know how to write better software and the general public should hang us if we continue to refuse to deliver it.
Never, ever connect appliances to the Internet. I thought this was common knowledge on HN. Also pull the telematics (cell modem) fuse to prevent your car from snooping on you and selling your location and driving habits to data brokers and insurance companies.
> [Your train] never comes on time. Fuel is more expensive. The roads are still congested with traffic. Your kids are (probably) doing worse at school. Food costs more. Houses cost more. Rent is higher. Buying a computer or a PS5 is more expensive. Politics is still full of mentally unstable people. The environment is getting worse. You will still die of heart disease or cancer. Food quality is worse. Your kids can't find work. Wealth inequality is accelerating.
None of this has to do with AI.
> But hey...on the flip side...a lot of people are rapidly building software (that nobody is using), and AI is solving mathematics problems (that nobody - including mathematicians - wanted it to solve)
Plain wrong on both counts.
> Imagine what a "country of geniuses in a datacenter" will be able to do Apparently...nothing.
Nonsequitor, probably based on a personal beef. Datacenters, at best, are understood to be for hosting more AI - not for people to somehow live in as an arcology of discovery.
While trickle down theory is dubious when it comes to economics, it's clear that it is valid when it comes to technology. Tech trickles down.
Obviously we get things in return for the social and economic fallout, carbon emissions, pollution, destruction of education, etc. that AI has provided us. I just haven't heard a compelling argument for why that is nearly enough to warrant this.
In American history, when previous shocks like this occurred (e.g. with the Great Depression), we needed massive economic programs like the New Deal to restore consumer demand. When such programs are not forthcoming while wealth inequalities keep increasing, one can look to other countries for what ends up happening; either the disenfranchisement of ordinary people over time or unrest and redistribution by force.
This reply summarizes the problem from a different angle.
Nevertheless, my counterpoint to
> If they actually had access to systems that could do "years of top professional work" across software, cyber, science, and math, they would not mostly be leaving to start more chatbot companies, agent frameworks, eval tools, and "AI for workflow" startups.
> They would be starting normal companies to disrupt the global economies of the world.
> Insurance companies. CAD companies. Compliance companies. Security auditing firms. Logistics software. Tax automation. ERP replacements. Drug discovery tooling. Industrial design. Legal ops. Accounting systems. Boring vertical SaaS with insane output per employee.
is:
Many people who work in such industries already know an insane amount of inefficiencies. Unluckily, they are not in positions and have no political power (with respect to organizational policy) in the company that they are capable of changing the situation.
Additionally, many of these sectors are heavily regulated, so it is very hard to make a difference in this sector as a startup founder, and even if there is an entrpreneur who founds a company to change the status quo, an insane amount of work in the company will be handling the big amount of red tape.
> Senior Director of Product in Insurance Tech
anakin_padme.jpg
Those LLMs are being adopted by the huge enterprise itself and being surrounded from all sides by a very complex procedural harness (aka regular programming) that stops the LLM from doing dumb stuff.
Videos are shittier. Most songs are even shittier than the autotuned pieces of shit already were. Ads are worse (even though nobody thought that was even possible). Customer support is worse.
> But hey...on the flip side...a lot of people are rapidly building software (that nobody is using)
That nobody is using and that are worse than what we used to have. And because nobody's using them, the countless issues plaguing them aren't even reported by users anymore (there are no users anymore). Full of bugs, usability issues and gigantic security holes (but which doesn't really matter because nobody's using them).
It's not even clear that it's making long-standing software (like say Linux or Emacs or Blender or QEMU) better at all.
Now I'd say though: most of the issues you mentioned have nothing to do with LLMs / AI. But it's very clear that AI ain't solving much of the world's problems and that the promises we got years ago didn't materialize.
Even if a certain small group of people are also using the hype surrounding this tech to attempt to destroy the home PC market and reverse the personal computer revolution. They will not succeed. There is more of us than them. Even with "thousands of frontier LLMs". A human+llm is going to be more powerful for a long time rather than a bunch of LLMs.
Everything isn't getting worse constantly. The data is quite clearly the opposite. How old are you?
Food has gone way up in cost and down in quality.
Private equity bought out the healthcare system where I lived so now it takes months to get an appointment and the quality of care is horrible.
Not that matters since the health insurance I pay hundreds for a month wouldn't cover anything anyways because it's through UHC.
My job keeps getting rid of positions because they want to "do more with less" but it just means I'm doing more work while being paid the same.
If I buy a new car it will spy on me and sell my data. The state DMV already sold my personal info without my permission and now I get constant spam.
Private equity bought out many of the restaurants I liked going to and ruined them.
Third spaces in my city have been mostly destroyed.
The social media sites I used to use were overrun by bots.
I can't upgrade my PC because hardware prices have gone insane.
Gas is like $5 a gallon now so I can't really justify going on any trips anymore.
Private equity bought up most of the rentals where I live and so now I pay much more for rent on top of crazy monthly fees while the building I live in is falling apart because they refuse to fix it. Housing prices have also gone up so much that I have no hope of owning a house.
And more.
AI has fixed none of these issues.
So basically you get personal digital assistant - very useful - but not changing my life.
What would I need for real transformation though? Food, house cleaning, commute. Even if we had ASI tomorrow, what could it do to change my life, besides taking away my job? Design good house robot to help me with food preparation and chores... Maybe finally self driving cars launched in my area? Or it will take control, "optimize humanity survival chance" and realize we don't need as many people on this planet.
(Those are all examples from just the past 4 days, and just the ones that made it into mainstream media.)
> self driving cars launched in my area
It sounds like the transformation would be a massive collapse of the economy with scores of people unemployed.
> Or it will take control, "optimize humanity survival chance" and realize we don't need as many people on this planet.
Maybe, but more likely it will be the humans who control the AIs and hold the power who will make this decision.
Unless you destroy their wealth now.
You’re implying that the deep lizard state planned the LLM craze?
Sama might be a lizard..
The Gartner Hype Cycle [0] is real, and ubiquitous. We are teetering somewhere near the top of the Peak of Inflated Expectations, about to plunge into the Trough of Disillusionment. A lot of very rich people are about to lose a lot of money, and we can only hope that they don't get bailed out (again).
Then, once the bubble and hype is over, the long, slow work of actually doing useful things with this new tech starts. That's when we'll see it affect everything else.
[0] https://en.wikipedia.org/wiki/Gartner_hype_cycle
I've been hearing this for a year and a half now. That word, "ubiquitous", is getting real mighty challenged. I have tried to insulate myself as best as I could and it just never happens and every other day I wonder if my foil hat needs adjusting. Does yours?
Regardless, people have been making your prediction for a couple years now
Even if all this is true (personally I doubt it), it assumes the bubble won’t burst before they’ve shifted the money from all our 401ks, IRAs and index funds into their own pockets.
I've done multiple life choices, away from AI, believing it to be a hype, but this hype is still growing, still not dying.
Both in 2019, as well as in 2023, i.e. not 2 or 3 months ago.
I, personally, think that choosing to move away from hyped tech is probably a mistake. The trick (imho) is to accept that it is hyped and will collapse, and use it while it makes sense to use it. Just don't bet everything on it. Have a backup ready.
When I asked why Musk doesn't put his money towards non-profits my Musk supporting friends say it's because he "doesn't believe anything but for-profit entities can solve the world's problems".
So why isn't he solving homelessness, famines, war, etc. with capitalism?
I’m not sold, but I’ve heard this line of thinking many times
As long as the world revolves around the central focus of making as much profit as possible as quickly and easily as possible the same cycle of enshittification will continue.
The question asked by decisionmakers is never "How can we leverage this to make people's lives better?"
It is always "How can we leverage this to make more money?"
And those two aims are almost always in direct conflict.
Companies are not trying to eliminate physical media because it is a benefit to people. They are doing it to drive growth for the sake of growth.
By comparison, we're in the early years of AI. We're all still figuring out what this thing even means.
However, the counterargument falls entirely flat. "Whatever benefits AI has to offer on a societal level are going to take a long long time to show up". We don't have the luxury of time because AI disrupting communities and livelihoods faster than perhaps any other technology we've ever seen, with the most capital that has ever been put into any tech since the railroads.
Even if you're optimistic we'll eventually get the cancer cures and fusion reactors, doesn't it make sense from the perspective of an ordinary American to regulate the harms like the homework cheating, parasocial relationships, job losses, cyber attacks, etc., especially if it will take years to get the benefits?
I don't see much evidence in the economic data that AI is disrupting livelihoods. Unemployment has been stable around 4% for a while and productivity growth has been at historically normal rates for the past 5 years. I think a more likely explanation is that AI is increasing uncertainty in some industries, which are seeing higher employee turnover, and many of those people are moving laterally into other jobs where their skills are more likely to be useful.
I don't think anyone in this thread has argued against mitigating harms. The OP was purely about current capabilities and the implications for technology in society.
Some of the most disruptive periods in history brought about the greatest long term improvements in society. This is the price we pay for changing how things are done, sometimes for better and sometimes for worse. We don't know which one it is for AI yet.
This is what is so frustrating about this whole ai thing is we aren’t learning from the mistakes of even the very recent past
I'm not sure why you think it was a massive mistake? The taxi industry was awful and unsafe, food delivery for many local places was either not available or poorly implemented.
Also many whole food chains didn't have proper delivery for groceries.
I feel similar about my robot vacuum. It actually takes a surprising amount of labor facilitating the damn thing. I have to walk through the house and basically vacuum proof it so it doesn't eat various cords, rugs, lifting chairs up and putting them someplace so it can get under, etc. Not sure its really a labor save to when I was just brooming the entire apartment in 20 mins and not having to move anything around due to reach of the broom and it not having a tendency to eat things. Hand vacuum I can be more carefull too about the fringed edges of rugs and don't have to fold them over themselves.
I still do it because I already bought it mostly.
e.g., my home is always vacuum-proofed. It automatically runs through all rooms except for the bedroom at night, finishing just before my alarm clock rings, then vacuuming the bedroom an hour after the last person's alarm clock has rung.
This way I never actually notice it, but the home is always perfectly clean as if by magic.
Similarly, if you use separate dishwasher detergent, antisurfactant and salt, and properly dial in your water hardness, it'll be able to clean everything no matter how dirty the dishes are. Including dried-on tomato sauce. Just put the dishes in, and they'll be clean the next day.
Of course it takes a bit of effort to set this up, but then it just works for a long time.
Similarly, I'm a big fan of handwriting software with care & focus on quality, which takes a bit longer at first, but then allows you to trust the software to continue running for years to come.
And then of course, all the cat's toys.
Sure, it's not perfect, but it's still really convenient.
This might be the next thing the robot vacuums all start working on. One has already released a first version: https://www.youtube.com/watch?v=GykaEgNkkVQ
there was never a need to rush other than for capitals self interest to control
if frontier labs wanna pace the frontier, they should be honest and return the compute back to the people
[1] https://sumsec.me/2026/liang-wenfeng-investor-meeting-transc...
i don't really think distilling is even that fundamental anymore (nor was the evidence ever that strong - https://msukhareva.substack.com/p/is-kimi-k3-really-distille...), the game is all RL now, and you don't need american progress for that at all (just compute and data, and the data brokers will sell to anyone no problem)
i do think there are black market proxies from china to the US trying to sell "claude" and "deepseek" access mixed together, and i do think that prompt distribution is probably going to be very nice datasets
Like, just at a bare minimum.
These are “use all the AI” companies with infinite resources, after all.
I don't understand why these big labs aren't doing more to improve the electricity situation. Dump a billion dollars worth of tokens into fusion analysis, research, and simulations and try to force a breakthrough. You could buy a lot of good will with unlimited cheap energy that liberates us from fossil fuel dependencies.
Fusion research and cancer research requires expensive experimentation in the real physical world. You can’t just throw a billion dollar worth of tokens to handle the problem.
Look at all the roadblocks for data centre buildouts which the industry certainly has a financial incentive to complete as fast as possible.
When the tokens are hooked up to physical devices (robotics of various sorts), then the tokens pay for themselves by being physical workers, or human wage replacement.
And a robocop will follow whatever is passed for law, damn the masses who might vote otherwise. And subjugation will be enforced by token, LLM rule, and machine. They have no union or ethics to complain to.
If you can make a GPU more efficient and cheaper than Nvidia does you can make a ton of money. Plenty of labs around the world are working hard on this problem and yet Nvidia is still dominating (and the market believes this isn't changing any time soon)
Same things for robotics. It's unlikely you'll have a vertical stack on all the different things needed to manufacture your robot on a massive scale. What happens if some geopolitical spat makes getting rare earth minerals and battery tech from China impossible?
Compare that to tokenmaxxing in programming and math, where the only thing you need is a gpu cluster and electricity.
> AI is solving mathematics problems (that nobody - including mathematicians - wanted it to solve)
Strange viewpoint. In mathematics any novel result is useful as it expands the scope of other results. Who are these mathematicians who didn't want AI to solve the problems it did? I see trepidation about the future of the field but that is not equivalent to not wanting them to be solved.
Some examples are finance people directly asking for API access now so that they can make their own tools. These tools used to be maintained by teams of developers.
Many business making their own customized scripts to make their tasks faster.
From an economic point of view, it seems like there are some productivity increases, but I don't think these can be concluded for a while. The rational thinking rarely account for second order effects. Although business leaders rarely care about second order effects, a worse business is not immediately noticeable, compared to hiring cuts and the money it saves
Again, strange viewpoint in the field of mathematics of all places, unlike, say, economics where doing something one shouldn't has disastrous real world consequences. Until it's discovered and proven, one doesn't know what would be valuable and what would not. I bet your sentence would not be said about humans trying to solve problems, no one would tell them they "shouldn't."
https://terrytao.wordpress.com/2026/10/07/ahm-statement-on-o...
A long time ago, another student of mathematics described it to me as climbing a mountain. No single step is hard, but you must make many steps to get to the point that no one else has made a step.
Trust me bro
Not for Tesla owners in select countries, where their car now drives them around 99% of the time, or the people of Austin Texas, that can now hail a fully driverless Cybercab for cheaper than an Uber!
>Food quality is worse.
Food quality is higher than ever if you want high-quality food!
Unfortunately, the people crave slop.
It might take a decade to take what we learned from AI and turn it into something that will benefit the masses.
Most of it can't be broken down into something as simplistic as a dishwasher.
Most people won't notice cancers being found faster due to AI being better at pattern recognition than a doctor, but it's still a great benefit.
Your pessimistic view doesn't hold up against the data. For a lot of the problems you mentioned, we're objectively in a better place now than almost any other time in history. Of course, things can always be improved and we should never grow complacent or accept injustices. But I would say in a lot of ways there's never been a better time in history to be alive.
Now it’s comments like this lamenting that the amazing LLMs we’ve had since early 2026 haven’t entirely revolutionized the whole physical world and even global geopolitics yet.
> But hey...on the flip side...a lot of people are rapidly building software (that nobody is using),
I guarantee a lot of software you use today has substantial AI development in recent releases.
Even the Linux kernel is getting a heavy input from AI and Linus is advocating for it. Unless you’ve frozen your entire software stack in 2024, you are using software with AI generated code.
> and AI is solving mathematics problems (that nobody - including mathematicians - wanted it to solve)
Saying nobody wants those problems solved is flat dishonest. The complaints from some mathematicians are that they and their peers didn’t get to put their names on the papers, not complaints that the problems were solved.
That's not fair. To first approximation, most of the progress in human condition of the last, like, forever is implementing solutions that people didn't see value in at the time of creation. Woz built a toy that could barely run software at all. The human genome project was panned as a boondoggle. Penicillin was a mistake.
The first fire pit probably looked like a dangerous stunt to the tribe faced with it, even.
Does that mean everything will come up roses? No. But LLMs only started talking to us for real like three years ago. Let's not just that fast.
Africa is going to look like wakanda in less than 5 years if WW3 gets canceled
Do people need permission from mathematicians to solve math problems? Is one of the Millennium problems not worth applying a new system to work on it?
I understand the despair people are feeling at AI automating their work, but we have to be serious about the fact that the frontier labs are just responding to natural incentives. Mathematics, coding, all are just applications of logic, and it makes sense that a system trained on inferring patterns from data would work well in that domain.
I'm pretty sure OpenAI's proofs, while difficult to understand, have a net-zero impact on human understanding of mathematics. It took the first Millenium problem years to verify its accuracy. It's only been days since OpenAI released its problems, surely it would take years to evaluate whether the solutions meaningfully advanced human understanding.
https://mathstodon.xyz/@tao/117395269325940185
https://mathstodon.xyz/@tao/117394287977270632
All mathematics researchers in the whole world have agreed upon that objective apparently?
Many modern medicines do not have understood mechanisms of action, yet we know they work. Are the people who found them not conducting drug research?
Do we even have a different term for what the AI labs are conducting if it's not mathematical research?
Additionally, I would argue that any work that adds to the total corpus of known facts increases human understanding, even if the path to get to each fact isn't understood. It's hard to argue we'd be better off with less total information.
What the AI labs are doing is mostly AI research. They are trying to test the limits of AI's ability to do mathematical research. This has been a productive subfield of generative AI research. So far, AI can't do mathematics research unaided - it can't produce both an original result and explain it to people. But they are getting better. Developing tools to aid mathematics research is not itself mathematics research, though it may both require new mathematics research and also contribute to mathematics research.
The difference is the subject area that is being developed. Developing the Lean theorem prover was mostly computer science.
1 : studious inquiry or examination especially : investigation or experimentation aimed at the discovery and interpretation of facts, revision of accepted theories or laws in the light of new facts, or practical application of such new or revised theories or laws
2 : the collecting of information about a particular subject
3 : careful or diligent search
Nothing here requires human understanding to be the goal. In fact, it explicitly calls out "practical applications" as one of the purposes that can be aimed at.
The whole field of applied mathematics is for the purpose of finding practical applications. Therefore any research conducted in this field is not for the exclusive purpose of human understanding or knowledge.
I'm not complaining, you're just restricting the definition of mathematics research in order to discredit what the AI labs have achieved.
Mathematicians are not gate keepers but I understand the frustration. If someone claims they spent 30 million dollars to solve one of the most difficult problems yet there’s no new knowledge, understanding, nor certainty. Then what’s the point besides marketing?
We've just locked out a large part of modernity because the shitty American auto industry is in swing states and also gender.
We are in that condition because people keep voting for these idiots, or not voting at all.
Just a crazy amount of doom and gloom all around this thread.
You gave your own example for the doom and gloom.
The fact that the means of this capability is in a few hands does suck. But I’m increasingly optimistic that it’s going to be commoditized through open weight models and performance improvements. There is enormous demand for innovations on these fronts.
At multiple jobs I've noted that I felt the pay should have been less. Not so that the employers could take a bigger slice of the pie (in fact, my preference is to be in an anarchist co-op, but I understand that's a wild take and might be terrible in practice). My hope was that we could filter for people that genuinely want to do the work and take on fewer people that were there for the money. At the same time more money is more better. And I like having money. It's complicated.
As I've said many times before, I didn't come to Silicon Valley to get paid. I came here to make computers do cool things alongside other people. I need income to do that, but I try to make sure I'm in a situation where the income isn't something I need to think about all of the time. If you're doing mercenary work that's not possible.
I think the amount of protectionism we’ll see could be absolutely enormous. Not only do incumbent corporations not want to lose share, incumbent countries won’t want to share either.
Coding has been transformed but none of those areas of our lives have been substantive transformed by AI yet. Whoever does, will create a lot of value for the world.
This is an indictment of mathematicians, not AI.
So just let them spend their money and they will eventually fail and bankrupt, right?
I would love it if we can just take the win on having created an amazing tool set for coders. Can’t this be enough?
If Americans cared about that they wouldn't have put the far right into power. I think its a little naive to expect AI to address any of those things when they are not in the realm of AI to address. These are political policy and Americans simply vote pro-capital owning class and anti-working class for misguided reasons.
In the workplace, AI can be quite a time saver so I think people getting no benefit is being disingenuous.
It's actually sad how much whinging doomers are doing and how you're all agreeing each other like some crazy death cult. I think you all want the world to got to hell just so you don't have to accept change.
> "...how much AI has either not improved life at all or actively made it worse from the perspective of the average person"
This is empirically, obviously and blatantly FALSE.
Claude and ChatGPT alone have 1.5 billion active monthly users. Do you REALLY think all those people are using it because it all sucks and is useless? That's 1 out of 8 people on the planet and it doesn't even count the million other AI companies out there (Google and Meta included).
Off the top of my head:
- Have you seen the TikTok videos from normal people talking about how Meta's Muse amazingly helped them do stuff like sell everything in their garage online with just a picture, or proactively helped them not miss appointments? I have.
- Have you talked to people who've used Claude to manage their small business or consulting by dealing with emails and rescheduling tasks? That's what my therapist friend is doing - all without any sort of technical help either, which is incredible.
- Have you been personally astounded at ChatGPT's ability to diagnose the problems with your car simply by your rambling description of a noise it's making? I have. It's really nice to go to a mechanic and have a general idea of what might be wrong.
- Have you talked to non-techies who solved their computer problem with the help of an AI chat giving them the right prompts to enter? It's so adorable how proud they are! Unlike getting help from people, AI is willing to walk them through their issues step by step, answer their dumb questions and not make them feel like they're stupid just because they're not technical. It's honestly heartening to see someone feel like they understand how their computer works for the first time in their life. (Actually, reading this, I realize it's an analog to me understanding my car for the first time. Heh.)
This stuff isn't even a smallest, tiniest, microscopic sliver of what AI is doing for people right now.
You're just being willfully ignorant or obtusely closed minded because you don't like change. Too bad, things are changing as they always do. Accept it.
I suspect the real world change is going to happen unbelievably rapidly and isn't far off. Right we don't quite have the hardware to embody AI in a generally useful way, but that's coming fast.
Within a couple of years arguably any task that requires just knowledge or intelligence is something you'd be better off giving to an AI than a human. And decent humanoid robots are maybe a decade behind that.
In the same way people growing up in the mid 1900s saw rapid changes in their way of life with planes and cars replaces boats and horses, we'll see something similar with robots replacing human labours. Right now we're still in the phase where people don't think humans will ever be able to control flying machines and that cars are reliable and slow. Things will change.
People didnt get a new car after they got. Internet in 2001. They didnt get new dishwashers.
The adoption of these technologies spurned ton of innovation in these spaces, and I feel like with LLMs its the first time I've seen genuine frustration from the industry directed towards users for not adopting the technology fast enough. This attitude did not exist in the 1980s, it didn't exist in the 90s, and it didn't exist in the 00s/early 10s either.
Whether you like it or not - technology causes compression of life cycles. You have a hit product? Great - I expect you to have crazy high growth rate. That’s the expectation of a world that is so deeply interconnected.
People need to be patient and put in the work to build great products, rather than lament its slower-than-desired (albeit still quite fast) adoption curve.
We can have unlimited new entertainment content for passive consumption. That's about it so far. 17,000 new Asian vertical dramas a day now.
AI won't show itself as a big new thing, but possibly in improvements in small little things
Most people won't use AI directly, except maybe as a smarter Google. In the same way people don't use HTTP directly
https://news.ycombinator.com/newsguidelines.html
The comment you're replying to is one of the handful that is actually useful, and the fact it was at the top of the thread until you hammered it down is proof everybody else thinks so too. But I guess we can't disturb the techochamber.
Seriously, just fuck you. There is a reason more and more people hate you as time goes by.
Did the writing also contain emotion? Yes. As all good persuasive writing should.
The main issue is that it doesn't engage with the substance of the post, and in fact, it exemplifies the very point the post is trying to make: there is a big and growing gap between those who recognize the benefits or potential of AI and those who don't (or who only see negative effects or negative trends in society).
I'll make it clear that I don't think your complaints or concerns are unfounded or invalid. There's merit and substance to all these issues, and it's fine to discuss them on HN. They are discussed all the time here, and they can and should continue to be discussed in the future.
But an important part of being the kind of curious commenter that we want everyone to aspire to be is that we can look at topics holistically. We need to be able to look at the opportunities and risks together and discuss ways that the benefits can be captured and shared throughout society, and the risks and costs adequately recognized and curtailed. Most people I know of in the tech industry are taking these matters seriously.
It's a vast and complex topic to discuss, and here's not the place for me to get further into it. But the point is that it's a bigger and more complex topic than your comment adequately recognizes, and we need people to always be making the effort to be thoughtful and curious rather than fulminatory or denunciatory in their participation here.
(Also: I can understand you may not easily recognize that your comment breaks the guidelines; it's common for people to not recognize that their words seem more inflammatory when read by others than they intended them to be.)
The people who are optimistic about AI believe it can help to address or reverse these trends. It's fine to debate the merits of that belief, but the original comment wasn't trying to do anything like that.
https://news.ycombinator.com/item?id=50055304#50060177
> It would be really cool if you guys enforced these rules uniformly.
I agree! But we can't, because there's way too much content for us to read it all, and we can't moderate what we don't see.
https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...
Can you point out what a good way of expressing disagreement or skepticism about a subject matter looks like on HN?
(My comment is actually snark, someone ban me off this hellhole of a site please)
There are so many comments on HN that would justify your attention with regard to this specific rule. This is not one of them.
E.g. some prominent folks definitely think LLM-based approaches will hit a wall, perhaps LeCun most notably. I also read a guest post on Terry Tao's site (which I really liked) that argues that, for all the very impressive recent AI results, they still operate within the "convex hull" of their training data: https://terrytao.wordpress.com/2026/09/13/happy-those-able-t...
I'm just curious if there is any actual data or evidence that takeoff (i.e. RSI, "the singularity", whatever you want to call it) is inevitable with current approaches.
But when you squint your eyes, LLMs are just a least-effort manifestation of the wider neural net direction we’ve been chipping away at for decades. I don’t give a fuck whether LLMs are here to stay; they’re proof that scaling neural nets gets you pretty far.
The real question is: will neural nets hit a wall? As a human, I certainly hope so. But I’d bet that they won’t. I’d bet that we’re igniting the ~~atmosphere~~ universe in slow motion.
But the "convex hull" idea always came across as wishful thinking and hand-wavy pseudo-science to me.
Here's a rebuttal [0] of it from Tim Gowers (uses another math metaphor, but still clear and the whole thread is worth a read):
>...the popularity of the idea that AI "merely explores the convex hull" mystifies me. Why wouldn't it be more like the subgroup generated by existing mathematical knowledge? (That's also a metaphor -- I'm not saying there's a group structure.)
>The convex hull metaphor suggests that AI isn't having radically new ideas and will eventually run out of things to do, whereas the group metaphor suggests the possibility of building up in small steps to something well beyond what you started with, with no obvious limit.
[0] https://twitter.com/wtgowers/status/2108090920864268679
We're in new grounds, no one knows which way it'll go.