The fact that the author is using PowerBI as the example makes me think they are completely oblivious of the actual impact and utility LLMs have in other fields such as programming. Claiming it's useless and that anything it outputs is "markedly inferior to that of a human doing the same task" is an absurd basis to use for their argument.
Share examples outside of tech. The simple: we're winning / their losing without context sounds cheeky but doesn't help anyone understand your perspective.
So what has an "LLM" truly improved in your life outside of simplistic automations/code generation/textual work? It's definitely not improved support. The only support interaction that I've noted to be somewhat better was one (of a few) with Amazon where the bot was able to refund a purchase without talking to a real person. But even then it was a case wherein a human would have probably done it just as quick. Is there a win there? Sure: for Amazon but it didn't "delight" me. And a chatbot is the lowest of low hanging fruit all the "optimists" claim an LLM can do so much better than humans or deterministic code.
How would you say LLMs are impacting the markets? Do you think they're useful to the economy thus far? Or are you looking at the spend of a small fraction of companies that could cause a broader recession when the cyclical funding dries up? If you're going to go to open weight models - well those don't appear out of thin air. Is your focus on the win/loss truly that myopic?
Again, your statement seems absurd to me. I think most people can agree there are useful modalities with LLMs. But it's always at a cost and I, like the author of the article, don't think that LLMs are as universally useful as many "optimists" believe. But to say that the pessimists can't actually rationalize their thoughts, when you provide none of your own, is truly foot in mouth garbage.
I do feel that a lot of pessimists I hear keep banging away about hallucinations and untrustworthy outputs. It makes me think they’re willfully ignorant of the progress since they tried ChatGPT 3 several years ago and were underwhelmed.
I’m happy to engage in discourse on the impact and risks to environment, society, jobs, mental health, etc. There’s plenty of room for thoughtful exploration there.
The one thing I have no patience for is that LLMs produce nothing of value or have no technical worth.
It's hard for me to take environmental risks regarding AI very seriously as well (and this is coming from someone who chooses to not drive or own a car, in large part for environmental reasons).
The recent efficiency gains in AI have been unparalleled in the history of computing as far as I know. We often point at how a single computer used to occupy an entire room, and over the course of decades of progress, are now wildly surpassed by computers that fit in our pocket. Inference of frontier language models that used to occupy entire data centers earlier this year (Claude Opus 4.6) are now comfortably surpassed in capability by models that can run on a single high end consumer grade gaming PC. I predict data center construction will collapse not just due to pushback from people who hate them, but also lack of demand from people who actually do use AI. People who like AI tend to like running their AI at home, and now they can.
The most important risks in AI right now are the social ones you mentioned. How does the economy evolve? How do we take advantage of this to maximize joy and love? Perhaps most importantly: how do we avoid (incentivize against) falling into a techno-dictatorship?
People who understand computers and those who do not.
I have to keep reminding myself though, many of those pessimists are people “getting paid to not understand” as the saying goes.
It took 30 seconds with that Davinci model to predict where we are today, where we’ll be next year, and the eventual destruction if not outright extinction of humanity as we know it. Maybe the future is so grim many are simply in the denial stage still.
It’s also just false. The good LLMs already produce output far better than the average programmer in most domains.
People like to talk about code as craft and art but the vast majority of the code that was written has always been mediocre at best. LLMs already beat it.
A focused experienced skilled programmer can do better but the market niche for that kind of code is actually small. Most commercial code just has to be good enough to accomplish a business need.
If you think of LLMs as statistically averaging machines and think of human programmer skill as a bell curve, it helps understand the different experiences with LLMs of different people. Those on the left hand side of the skill curve will see an uplift, those on the right hand side will experience degradation.
The more pertinent question in my mind is.. why does it still matter? Your customers don’t care about elegant code, they care about working products. There is some relation to good code and good customer experience e.g in terms of performance, but in general your customers won’t feel the impact from poor code. That assumes both pieces of code are functional though, and you do need to babysit LLMs to make sure it actually remains functional.
On the last point: your customers absolutely feel the impact of your poor code. I’ve worked on very well written codebases and very poorly written ones. The well written ones have had far fewer outages and complaints. When you have multi million line codebases, being “functional” (I understood you to mean this as “working correctly” rather than the technical term) is a spectrum.
> If you think of LLMs as statistically averaging machines and think of human programmer skill as a bell curve ...
The problem with this view is that in the scale of LLM progress, this is essentially accurate understanding of archaic models. Modern models are not trained to program using human generated code. They are trained in agentic harnesses where they are able to access real computing environments, and they are rewarded for completing tasks within that environment. This is in contrast the being rewarded for the most likely next token, which used to create essentially an average of human output like you say. Modern training frameworks are able to comfortably produce models that perform well above the center of the bell curve.
90% of what even the best of the best programmers do can be accelerated by LLMs. It's not just coding, but research and code archaeology, bouncing ideas around, automated testing etc.
> A focused experienced skilled programmer can do better but the market niche for that kind of code is actually small.
At this point I think the only reason someone would do better without an LLM is an unwillingness to use it effectively. Even if you’re going to write all code by hand and only use an agent for validation and debugging, that’s a significant boost.
Every day you all just find your little articles and blogs on here to get mad at and then write the same exact response to it.
If you really are on the winning/right/good side of this debate, then just be in it, you don't need to sweat it all so much. We have all heard what you or anyone else are going to say on the matter, trust me.
Why care at this point? Why even give the conversation any air by going and responding with the same lines over and over? Is it really just too much to y'all still that some guy has the wrong opinion on the internet?
"The first point is the question of whether an LLM can do work not markedly inferior to that of a human doing the same task or not. This one is debatable. An LLM absolutely cannot produce work to the highest human standards, or even a good human standard at the moment, and it's unlikely that it ever will."
This could have been fine maybe two years ago, but saying that in the era when LLMs are solving Millenium Prize Problems is incredibly obnoxious. All those opinionated "AI articles" feel like ragebait at this point.
GPT-4 felt like a slightly dim but still very useful intern.
GPT-6.1 (and Opus 5.5, etc.,) is a solid mid-career pro who can produce outputs very much on par with "good human standards" -- and, in some cases, far better than that.
Whatever in-house dev model they're using at OAI/Anthropic is, by all accounts, legitimately superhuman and capable of solving outstanding technical and mathematical problems with three hours of "Pro thinking."
OP's complaint is foolish, it reads like somebody who has never even used these things and has no idea what their capabilities actually are.
> who can produce outputs very much on par with "good human standards"
“Can produce”. Is doing a lot of heavy lifting there. It can, but then they also regularly write truly bafflingly bad code that no decent experienced dev would ever produce.
I’ll give you a few examples from last week. I had claude whip up a Kafka consumer. After a lot of back and forth I was pretty confident it worked and then I decided to dig into the code. I spotted some real head scratchers.
When starting up the consumer it wanted to pass a specific rebalance strategy, but then it realized that the library it was using didn’t support that strategy, so it needed to use the default.
Instead of just deleting the config option (and maybe leaving a comment), it created a function that removed that key from the map and used it like:
Then I started looking at the tests. They looked good but then I realized it wanted to be able to test without Kafka so what it had done was make multiple public functions that were mostly duplicates of the real functions
except that they didn’t call the Kafka library. Then it just tested those instead.
Another frontier model reviewer was supposed to prevent that kind of thing but it didn’t catch it either.
I had 10 public functions that were only called in tests throughout the whole project. Essentially the tests were useless because they weren’t actually testing anything.
Clearly current frontier models are useful. And maybe they produce code that is better than the median dev by many metrics. But I don’t think they are anywhere near consistently producing better quality than a say a 70the percentile dev with 10 years experience.
And even when the quality is better than the median dev, I’m finding that the bugs are not the kind of thing that humans or LLMs are good at spotting.
"OP's complaint is foolish" - I just shared what I thought was an interesting read. That doesn't mean that I hold the same views as the author of the linked post.
> An LLM absolutely cannot produce work to the highest human standards
yep. This is the choice quote.
Even most humans cannot produce work to the highest human standards. The fact is, most work does not need to be done to the highest human standards, it simply needs to be done to the lowest possible standard that someone would pay for, at the cheapest price.
The consumer chooses the cheapest price. Look at sweat shops, look at chinese manufacturing, look at offshoring. All of these produces an inferior product, and yet they are what gets the most profit and overwhelmingly the largest market share of consumer purchases.
yep. This is the choice quote.
Even most humans cannot produce work to the highest human standards. The fact is, most work does not need to be done to the highest human standards, it simply needs to be done to the lowest possible standard that someone would pay for, at the cheapest price.
TFA:
The first point is the question of whether an LLM can do work not markedly inferior to that of a human doing the same task or not. This one is debatable. An LLM absolutely cannot produce work to the highest human standards, or even a good human standard at the moment, and it's unlikely that it ever will. Most human work, however, is not done at a good standard. At best it's mediocre, and at worst it's downright horrible. The bulk of human knowledge work done in an awful of settings, then, is unfortunately more or less interchangeable with that of an LLM (seriously, so much human work is just awful: I wish I could say otherwise). On this point, then, LLMs probably pass the criterion.
I'd recommend that you take 15->30 minutes and read the entire essay. It's a good read.
This is the thing: they probably AI summarized it. Because in today's world: why read for yourself and enjoy the nuance of a human generated work? Most of the comments in this thread are, unfortunately defending the thing that will make them less useful and less relevant in the context that they want to "win" this argument. It's going to be a wild ride down. Enjoy your electrolytes.
Given the consistency in argument of all but one or two of the top-level comments posted at the time I commented, I find it likely that the commenters are some combination of bots and underpaid astroturfers working in very low cost-of-living parts of the world.
Also, I'd correct your opening sentence like so:
> This is the thing: they probably AI "summarized" it.
I've seen many "summaries" that have come out of those lossy compressors... when tasked with compressing inputs of any appreciable complexity, the output is often too poor to serve as a summary. What you get sure is summary-shaped, though.
> Most of the comments in this thread are, unfortunately defending the thing that will make them less useful and less relevant...
If your claim here is that most of the comments so far are astroturfers or similar, ignore the rest of this line. If it isn't, I'd recommend that you read TFA in its entirety... or go back and read it more carefully if you already did. A sizeable section of the essay talks about what I think you're gesturing at.
> If your claim here is that most of the comments so far are astroturfers or similar, ignore the rest of this line. If it isn't, I'd recommend that you read TFA in its entirety... or go back and read it more carefully if you already did. A sizeable section of the essay talks about what I think you're gesturing at.
100% in alignment. It almost seems like this is AI boosters looking to set the conversational agenda early to sway any dissent from happening. Real conversations are becoming harder and harder to have.
> Real conversations are becoming harder and harder to have.
Eyuuup. Even on The Red Site (which usually has fairly sane commentary... as long as simonw stays out of the conversation thread), the first commenters on the submission about this essay have very obviously either not read the entire essay, or are pointedly ignoring the parts of it that either conflict with or directly address the claims in their comments.
As is sometimes said by idiot Twitch streamers: Fuck my chungus life, I guess.
> commenters are some combination of bots and underpaid astroturfers
Always these baseless accusations. I've been here almost as long as you have and find it extremely offensive, just because I'm rebuking a flawed argument with basic facts. There's no conspiracy and you're breaking the HN rules by insinuating it.
Wow, this was ejected off of the front page. It's currently sitting at page 10 two hours after posting. Based on what I've seen over the years, carefully-written criticism about the products of the LLM industry rarely does do well here... for whatever reason.
Last I heard they had only data mined their own (human) users, submitted a pile of slop for (human) mathematicians to verify, and then withdrew much of it when it proved not to be a solution.
It's so tiresome. I'm going to stop getting into these threads altogether soon. There's nothing to discuss with the ones who treat it like a religious belief. I keep thinking that they'll just try some of the tools and see reason, but you can't rationalize with the irrational.
I wish people who felt this way would just go use the tools they hate so much before passing judgement. I think many of them used it once, in a much worse form, many years ago and not since. Judging these things by the standards of 3 months ago is unfair, let alone the standards of 3 years ago.
We now have professional mathematicians that are on the record carelessly dismissing major math questions being decided. And I'm not talking about having a subtle discussion about what proofs mean.
So it's not really about trying it out, I think. It's something more emotional.
I'll just point at this essay from five months ago and hope you folks read it carefully. [0]
I'll include this excerpt from the essay as an attempt to shame folks who feel the urge to continue to claim that the author is critical of "AI" because they hate the tech and never ever have anything good to say about it:
> I will say that Claude Code is a noticeable step change over the Opencode-based GLM models that I tried a few months ago. Those were bad enough to not be usable for anything serious or of real complexity. Claude Code, by contrast, can do some relatively complex things: it was able to perform the bulk of my website migration from Nuxt 3 to Nuxt 4 and to the newest version of Nuxt-Content for instance, which the GLM models had utterly barfed on. It wasn't perfect: I had to do quite a bit of tinkering after the fact to make things work, but unlike the GLM attempt, Claude did actually save me some time. Claude also did pretty well with adding Schema.org markup to my website and performing a number of other stupid-but-necessary SEO-related tasks on the same website: none of these tasks are really groundbreaking, but they're the kind of relatively complex pain-in-the-ass task that you'd hope we could automate: at this point in my practice, there's little of value that I can glean from doing those tasks by hand. If it's possible to make this kind of tool accessible to people at an acceptable, sustainable cost and without burning too many resources, then, I think that'd be a good thing.
> Claude Code also does quite a bit better than Opencode tools did at CI/CD tasks, IaC in OpenTofu, building Dockerfiles and that kind of task. It's still something I'm quite iffy on, but unlike the GLM models I tried earlier, what Claude Code produces basically works, at least initially. This probably makes it a lot more immediately useful than the GLM models were, as this kind of task is a large part of what tech workers actually do day-to-day, and it's the kind of task that a software engineer might have to perform while not having a very good grasp of the underlying technology.
I recommend folks take 15->30 minutes to actually read the essay, rather than either skimming it or having some LLM "summarize" it. There's a lot of subtlety and nuance that the current generation of lossy compressors is incapable of reliably compacting.
> The technology can, in volume of adequate output or in quality of output, overwhelmingly outclass what alternative technologies can do. A moderate improvement isn't really enough to force inevitability: it has to be a massive difference.
> Now we come to the question of whether or not the difference that LLMs bring to the table is sufficiently overwhelming to make their adoption inevitable, and on this I think the answer is a clear no. As established above, LLMs just can't do top-flight work at all: at the heights of basically any discipline, we're still very much relying on unaugmented human work.
The clear flaw here is that cost isn't taken into account. A new technology needs to be better than the status quo at a given cost/price point.
Examples: mass produced textiles, photography, IKEA. Ostensibly not the highest quality, but the cheapest at a price point.
This was published today, I had to check the date to be sure. I used to be an applied researcher working in big tech and for the things I wanted to build if I had to pick between LLMs and hand picking a team of 100 from the best programmers / researchers in the world I would pick the LLMs. Of course the combination would be even better. I've had LLMs immediately understand concepts that took me months to effectively communicate to a colleague and there are some concepts I was unable to communicate to any other human at all. Then the LLMs could build what I needed in less time it would take me to set up a meeting about it.
> Any actor that's at all serious in their field is, sooner or later, going to hit a point where LLMs cease being useful to them and actively hinder their further progress, and we absolutely do not see the massively overwhelming advantages that LLMs would need to have in order to become inevitable.
I can only conclude from this that I am not at all serious in my field. That's okay. I'm too busy enjoying myself.
If you don’t use AI to scan for security bugs yourself, someone else will likely be using AI to find them first. So in that way, at least, LLMs could be considered inevitable.
The first part of the article is a thoughtful framework. The conclusion that LLMs are not inevitable because they aren't useful or that their utility can be substituted by humans is an extraordinarily flimsy claim.
Human intellectual effort is the tightrope walker in Thus Spoke Zarathustra; AI is the jester. LLMs can already solve programming and mathematical problems better and faster than humans. It's only a matter of time before other intellectual disciplines fall. We would be profoundly stupid not to embrace this change.
Empathy and big-picture thinking over myopic specialization will be the new moat.
I read "The level of social opprobrium directed at LLM use and at people attempting to pass off generated slop as work is immense" and immediately checked for the author's Bluesky link. This person is living in a bubble. Most people in tech are now using LLMs at work and there is no social opprobrium discouraging them.
The rest of the analysis seems to take it for granted that we have hit a wall and LLMs will improve no further. OK. We'll see how that prediction plays out.
> Most people in tech are now using LLMs at work and there is no social opprobrium discouraging them.
I don't know if "most people" is true, yet. I do think that the opprobrium has been silenced, and in corporate environments that has been systematic and top-down. Tokenmaxxing, leaderboards, monitoring: this is all part of making sure developers do not spend time considering the ethics, because the conclusions have been rendered moot.
"90% of professional developers were using AI coding agents at work at least weekly in one form or another (local agents or remote cloud agents), with 68% using them daily."
Re. context:
"The Developer Ecosystem and AI Pulse surveys are localized into eight languages: English, Spanish, Chinese, Japanese, Korean, German, French, and Portuguese. We apply quotas on the required number of responses by region to help achieve accurate global representation. ... The Developer Ecosystem Survey has been statistically reweighted to better represent the global developer population by region, employment status, programming language, and familiarity with JetBrains products"
Fully unclear who, actually, is being surveyed. Developers generally through some industry-wide survey? Or JetBrains users?
(The survey was run on Jetbrains instagram page, apparently. I mean… that is a bit of selection and narrowing going on there)
Notice how half of them have heard about Jetbrains!? I don't know any developers I would be confident know about Jetbrains (even among AI-using developers)
(Also the graphs don't really support the idea that 90% are using them meaningfully for software development.)
We need to start talking about anti-AI psychosis. That people actually think their delusions fueled by these echo chambers are real is almost as worrying as the AI booster bubbles.
I'm gonna go ahead and make the prediction that this eventually morphs into a kind of religion, like flat earthers.
Nothing that AI will ever be valuable, everything AI does is steal, nothing it creates is new, everything that the big labs publishes is a scam or hype. Oh and "it's just doing X", where you can substitute "X" with some kind of surface level description of how the output layer in an LLM works.
These tenets are already pretty much formed, all it needs is to be formalized by some kind of name or manifesto similar to Q-anon, and we've arrived at a durable state of permanent denial.
There‘s hardly anything to talk about, since most of those anti-AI people use LLMs anyway. It‘s hypocrisy at best. If only that energy could be channeled in doing more about open-weight models-we need more AI-ownership, not less AI. That LLMs will somehow magically disappear should Anthropic or OpenAI go bust, is nothing but wishful thinking.
One thing I've seen with LLMs that I've not seen with any other emergent technology is the level of backlash against it. I think we're seeing a point of polarization in society where there will be two truly opposing sides that are against each other and no reconciliation will be possible between them (pro and anti-AI), and that it has a good chance of fundamentally destabilizing society.
> I've not seen with any other emergent technology is the level of backlash against it.
i think those who had backlashes against machines and automation in the past would be similar tbh.
It's because when people feel, even if remotely, their livelihood going down the drain, they do anything and everything to fight it (even if it would be against the interest of overall progress - nobody wants to be the "sacrifice").
I mean, Luddites were hanged for murdering people. We're not quite there yet.
They were not just fighting for their livelihoods, though; the fight was at least as much about dignity and fair treatment.
And they were correct to do so: skilled workers who lost their work and ended up in industrial employment did not have better lives. It took a century to a century-and-a-half for industrial quality of life to get back to pre-industrial quality of life.
Loss of dignity is something that the AI industry is not considering as it renders entire swathes of the economy button-pushers for chatbots.
When loss of livelihood appears it will be too late and at too great a scale to fight back in the workplace; the fightback will be somewhere else.
H.G Wells Time Machine has future society split into to: the childish consumer class and the driven engineering class, and the consumer class is the food of the engineering class. I could see this happening, with how cruel so many are, and trying to institutionalize that attitude.
I've seen a lot of people not wanting to learn to use a computer ("I prefer the old way") or not wanting to have a mobile phone ("I don't want to be reachable all the time") or a smartphone with touchscreen ("This feels like a too close interaction between the human and the device").
I haven't seen much splitting into two groups of people who are for and against. It's more that individuals can see upsides - does cool stuff, and downsides - something like a rival lifeform that will be smarter than us.
I think that's a pretty odd framing, because the upsides are pretty relative and lopsided to those receiving certain benefits. The people who are anti-AI often are those who can acknowledge that AI does cool stuff, but that it's not worth it and doesn't provide any real benefit except by helping people keep up in a frenetic arms race to develop new tech.
Personally, I'm against AI but even if it could never be smarter than us and if it were at the state that it was last year, I'd still be against it due to the environmental cost.
I'm mildly pro but agree on the environmental costs. I think they should try to make it green energy only though not much chance with the current US admin. I'm not so enthused with current AI. I use last years like language translation a lot and I'm interested in the possible sci-fi future like curing diseases, some sort of immortality, abundance etc.
I keep seeing it and it's really bipolar. Boosters think it's the best thing ever and downplays all the issues while antis think it's completely useless and a disaster environmentally and morally.
Good luck saying LLMs are useful but still a net negative in the filter bubble this author is in. No, you also need to think it has zero utility or you're the enemy.
One datapoint on the same people seeing pro and anti is Nick Bostrom (Swedish philosopher & writer) who was one of the instigators of AI doom with his 2014 book:
Well yeah, that's because it's political. The theat of job (and more importantly income) losses as well as twitter accelerationists, AI art slop spammers and anxiety trolls have really fueled the fire to the degree that people hate this stuff.
Not to mention how creepy Sam Altman and Dario Amodei are.
The author's absurd fantasy of how an LLM has to be used in order to fit their imagined adversary's idea of inevitable is absolute drivel. This article doesn't even build a straw man. It's just a pile of straw and manure.
“That surveillance capitalism is a logic in action and not a technology is a vital point because surveillance capitalists want us to think that their practices are inevitable expressions of the technologies they employ. For example, in 2009 the public first became aware that Google maintains our search histories indefinitely: data that are available as raw-material supplies are also available to intelligence and law-enforcement agencies. When questioned about these practices, the corporation’s former CEO Eric Schmidt mused, ‘The reality is that search engines including Google do retain this information for some time.’
In truth, search engines do not retain, but surveillance capitalism does. Schmidt’s statement is a classic of misdirection that bewilders the public by conflating commercial imperatives and technological necessity.”
Shoshanna Zuboff, “The Age of Surveillance Capitalism”
> In papers filed with the court on November 20, the DOJ told the court that Windows 95 would function without Internet Explorer 3.0, and that Microsoft should be required to offer computer manufacturers (OEMs) the option of installing Windows, free of Internet Explorer 3.0 in
“all” respects. In response, Microsoft explained that the latest versions of Windows 95 would not work under such a scenario
It helps if the author does not confuse what LLMs actually do with the general misunderstanding that "LLMs do work". They do not "do work" and have never "done work", and this misunderstanding is all over tech.
What do LLMs to? You know this: they are sentence completion engines, that through this process of sentence completion can be used to create software than then "does work". This indirection is critical, it is "how this works" and is the true power of LLMs that is STILL not being recognized. The power of LLMs is their communications capacity, their ability to complete sentences, which is so subtle the entire world of software engineers and artificial intelligence everybody has not identified that simple "variable expansion" as the true power of LLMs. LLMs do not do work, they enable the creation of software that then does work.
With a technology so subtle, want to know what is inevitable? A bifurcation of the population into those that believe in magic and those that understand it well enough to make magic nobody considered possible. Why did I say "magic"? Because people are giving up understanding, and that makes what we do magic.
The fact that the author is using PowerBI as the example makes me think they are completely oblivious of the actual impact and utility LLMs have in other fields such as programming. Claiming it's useless and that anything it outputs is "markedly inferior to that of a human doing the same task" is an absurd basis to use for their argument.
I know it's a typo, but I wonder if "programmering" could be a useful new word for the act of creating programmers!
Oops, fixed but I think you're onto something!
There are LLM optimists and LLM pessimists.
The optimists have been more correct in predicting the capabilities and usefulness of LLMs.
The pessimists seem to have a hard time rationalizing what they thought with what’s actually happening.
Share examples outside of tech. The simple: we're winning / their losing without context sounds cheeky but doesn't help anyone understand your perspective.
So what has an "LLM" truly improved in your life outside of simplistic automations/code generation/textual work? It's definitely not improved support. The only support interaction that I've noted to be somewhat better was one (of a few) with Amazon where the bot was able to refund a purchase without talking to a real person. But even then it was a case wherein a human would have probably done it just as quick. Is there a win there? Sure: for Amazon but it didn't "delight" me. And a chatbot is the lowest of low hanging fruit all the "optimists" claim an LLM can do so much better than humans or deterministic code.
How would you say LLMs are impacting the markets? Do you think they're useful to the economy thus far? Or are you looking at the spend of a small fraction of companies that could cause a broader recession when the cyclical funding dries up? If you're going to go to open weight models - well those don't appear out of thin air. Is your focus on the win/loss truly that myopic?
Again, your statement seems absurd to me. I think most people can agree there are useful modalities with LLMs. But it's always at a cost and I, like the author of the article, don't think that LLMs are as universally useful as many "optimists" believe. But to say that the pessimists can't actually rationalize their thoughts, when you provide none of your own, is truly foot in mouth garbage.
I do feel that a lot of pessimists I hear keep banging away about hallucinations and untrustworthy outputs. It makes me think they’re willfully ignorant of the progress since they tried ChatGPT 3 several years ago and were underwhelmed.
I’m happy to engage in discourse on the impact and risks to environment, society, jobs, mental health, etc. There’s plenty of room for thoughtful exploration there.
The one thing I have no patience for is that LLMs produce nothing of value or have no technical worth.
It's hard for me to take environmental risks regarding AI very seriously as well (and this is coming from someone who chooses to not drive or own a car, in large part for environmental reasons).
The recent efficiency gains in AI have been unparalleled in the history of computing as far as I know. We often point at how a single computer used to occupy an entire room, and over the course of decades of progress, are now wildly surpassed by computers that fit in our pocket. Inference of frontier language models that used to occupy entire data centers earlier this year (Claude Opus 4.6) are now comfortably surpassed in capability by models that can run on a single high end consumer grade gaming PC. I predict data center construction will collapse not just due to pushback from people who hate them, but also lack of demand from people who actually do use AI. People who like AI tend to like running their AI at home, and now they can.
The most important risks in AI right now are the social ones you mentioned. How does the economy evolve? How do we take advantage of this to maximize joy and love? Perhaps most importantly: how do we avoid (incentivize against) falling into a techno-dictatorship?
People who understand computers and those who do not.
I have to keep reminding myself though, many of those pessimists are people “getting paid to not understand” as the saying goes.
It took 30 seconds with that Davinci model to predict where we are today, where we’ll be next year, and the eventual destruction if not outright extinction of humanity as we know it. Maybe the future is so grim many are simply in the denial stage still.
[dead]
It’s also just false. The good LLMs already produce output far better than the average programmer in most domains.
People like to talk about code as craft and art but the vast majority of the code that was written has always been mediocre at best. LLMs already beat it.
A focused experienced skilled programmer can do better but the market niche for that kind of code is actually small. Most commercial code just has to be good enough to accomplish a business need.
If you think of LLMs as statistically averaging machines and think of human programmer skill as a bell curve, it helps understand the different experiences with LLMs of different people. Those on the left hand side of the skill curve will see an uplift, those on the right hand side will experience degradation.
The more pertinent question in my mind is.. why does it still matter? Your customers don’t care about elegant code, they care about working products. There is some relation to good code and good customer experience e.g in terms of performance, but in general your customers won’t feel the impact from poor code. That assumes both pieces of code are functional though, and you do need to babysit LLMs to make sure it actually remains functional.
On the last point: your customers absolutely feel the impact of your poor code. I’ve worked on very well written codebases and very poorly written ones. The well written ones have had far fewer outages and complaints. When you have multi million line codebases, being “functional” (I understood you to mean this as “working correctly” rather than the technical term) is a spectrum.
> If you think of LLMs as statistically averaging machines and think of human programmer skill as a bell curve ...
The problem with this view is that in the scale of LLM progress, this is essentially accurate understanding of archaic models. Modern models are not trained to program using human generated code. They are trained in agentic harnesses where they are able to access real computing environments, and they are rewarded for completing tasks within that environment. This is in contrast the being rewarded for the most likely next token, which used to create essentially an average of human output like you say. Modern training frameworks are able to comfortably produce models that perform well above the center of the bell curve.
90% of what even the best of the best programmers do can be accelerated by LLMs. It's not just coding, but research and code archaeology, bouncing ideas around, automated testing etc.
Automated testing is insane. LLMs are amazing at stamping out unit tests, running them, investigating failures, etc.
> A focused experienced skilled programmer can do better but the market niche for that kind of code is actually small.
At this point I think the only reason someone would do better without an LLM is an unwillingness to use it effectively. Even if you’re going to write all code by hand and only use an agent for validation and debugging, that’s a significant boost.
Every day you all just find your little articles and blogs on here to get mad at and then write the same exact response to it.
If you really are on the winning/right/good side of this debate, then just be in it, you don't need to sweat it all so much. We have all heard what you or anyone else are going to say on the matter, trust me.
Why care at this point? Why even give the conversation any air by going and responding with the same lines over and over? Is it really just too much to y'all still that some guy has the wrong opinion on the internet?
[dead]
"The first point is the question of whether an LLM can do work not markedly inferior to that of a human doing the same task or not. This one is debatable. An LLM absolutely cannot produce work to the highest human standards, or even a good human standard at the moment, and it's unlikely that it ever will."
This could have been fine maybe two years ago, but saying that in the era when LLMs are solving Millenium Prize Problems is incredibly obnoxious. All those opinionated "AI articles" feel like ragebait at this point.
GPT-4 felt like a slightly dim but still very useful intern.
GPT-6.1 (and Opus 5.5, etc.,) is a solid mid-career pro who can produce outputs very much on par with "good human standards" -- and, in some cases, far better than that.
Whatever in-house dev model they're using at OAI/Anthropic is, by all accounts, legitimately superhuman and capable of solving outstanding technical and mathematical problems with three hours of "Pro thinking."
OP's complaint is foolish, it reads like somebody who has never even used these things and has no idea what their capabilities actually are.
> who can produce outputs very much on par with "good human standards"
“Can produce”. Is doing a lot of heavy lifting there. It can, but then they also regularly write truly bafflingly bad code that no decent experienced dev would ever produce.
I’ll give you a few examples from last week. I had claude whip up a Kafka consumer. After a lot of back and forth I was pretty confident it worked and then I decided to dig into the code. I spotted some real head scratchers.
When starting up the consumer it wanted to pass a specific rebalance strategy, but then it realized that the library it was using didn’t support that strategy, so it needed to use the default.
Instead of just deleting the config option (and maybe leaving a comment), it created a function that removed that key from the map and used it like:
Then I started looking at the tests. They looked good but then I realized it wanted to be able to test without Kafka so what it had done was make multiple public functions that were mostly duplicates of the real functions except that they didn’t call the Kafka library. Then it just tested those instead.Another frontier model reviewer was supposed to prevent that kind of thing but it didn’t catch it either.
I had 10 public functions that were only called in tests throughout the whole project. Essentially the tests were useless because they weren’t actually testing anything.
Clearly current frontier models are useful. And maybe they produce code that is better than the median dev by many metrics. But I don’t think they are anywhere near consistently producing better quality than a say a 70the percentile dev with 10 years experience.
And even when the quality is better than the median dev, I’m finding that the bugs are not the kind of thing that humans or LLMs are good at spotting.
I don't doubt it can, sometimes, produce output far better than most human can. The problem is: what happens what it don't ...
It's likely similar to what happens when the humans don't.
In addition to making very different kinds of bugs than humans tend to, humans can’t deploy daily 20k LOC changes.
"OP's complaint is foolish" - I just shared what I thought was an interesting read. That doesn't mean that I hold the same views as the author of the linked post.
Very sorry, I meant "article in OP's complaint..."
And actually the article was quite interesting, though I believe it to be wrong.
> An LLM absolutely cannot produce work to the highest human standards
yep. This is the choice quote.
Even most humans cannot produce work to the highest human standards. The fact is, most work does not need to be done to the highest human standards, it simply needs to be done to the lowest possible standard that someone would pay for, at the cheapest price.
The consumer chooses the cheapest price. Look at sweat shops, look at chinese manufacturing, look at offshoring. All of these produces an inferior product, and yet they are what gets the most profit and overwhelmingly the largest market share of consumer purchases.
You:
TFA: I'd recommend that you take 15->30 minutes and read the entire essay. It's a good read.This is the thing: they probably AI summarized it. Because in today's world: why read for yourself and enjoy the nuance of a human generated work? Most of the comments in this thread are, unfortunately defending the thing that will make them less useful and less relevant in the context that they want to "win" this argument. It's going to be a wild ride down. Enjoy your electrolytes.
Given the consistency in argument of all but one or two of the top-level comments posted at the time I commented, I find it likely that the commenters are some combination of bots and underpaid astroturfers working in very low cost-of-living parts of the world.
Also, I'd correct your opening sentence like so:
> This is the thing: they probably AI "summarized" it.
I've seen many "summaries" that have come out of those lossy compressors... when tasked with compressing inputs of any appreciable complexity, the output is often too poor to serve as a summary. What you get sure is summary-shaped, though.
> Most of the comments in this thread are, unfortunately defending the thing that will make them less useful and less relevant...
If your claim here is that most of the comments so far are astroturfers or similar, ignore the rest of this line. If it isn't, I'd recommend that you read TFA in its entirety... or go back and read it more carefully if you already did. A sizeable section of the essay talks about what I think you're gesturing at.
> If your claim here is that most of the comments so far are astroturfers or similar, ignore the rest of this line. If it isn't, I'd recommend that you read TFA in its entirety... or go back and read it more carefully if you already did. A sizeable section of the essay talks about what I think you're gesturing at.
100% in alignment. It almost seems like this is AI boosters looking to set the conversational agenda early to sway any dissent from happening. Real conversations are becoming harder and harder to have.
> Real conversations are becoming harder and harder to have.
Eyuuup. Even on The Red Site (which usually has fairly sane commentary... as long as simonw stays out of the conversation thread), the first commenters on the submission about this essay have very obviously either not read the entire essay, or are pointedly ignoring the parts of it that either conflict with or directly address the claims in their comments.
As is sometimes said by idiot Twitch streamers: Fuck my chungus life, I guess.
> commenters are some combination of bots and underpaid astroturfers
Always these baseless accusations. I've been here almost as long as you have and find it extremely offensive, just because I'm rebuking a flawed argument with basic facts. There's no conspiracy and you're breaking the HN rules by insinuating it.
> There's no conspiracy...
I agree!
> ...and you're breaking the HN rules by insinuating it.
It's a good thing that I'm not, then.
> ...just because I'm rebuking a flawed argument with basic facts.
cough
Also, some meta-commentary:
Wow, this was ejected off of the front page. It's currently sitting at page 10 two hours after posting. Based on what I've seen over the years, carefully-written criticism about the products of the LLM industry rarely does do well here... for whatever reason.
I wonder why... o_O
Last I heard they had only data mined their own (human) users, submitted a pile of slop for (human) mathematicians to verify, and then withdrew much of it when it proved not to be a solution.
Maybe you heard wrong? Who knows... Ask your neighbor, if not AI! :)
It's so tiresome. I'm going to stop getting into these threads altogether soon. There's nothing to discuss with the ones who treat it like a religious belief. I keep thinking that they'll just try some of the tools and see reason, but you can't rationalize with the irrational.
I wish people who felt this way would just go use the tools they hate so much before passing judgement. I think many of them used it once, in a much worse form, many years ago and not since. Judging these things by the standards of 3 months ago is unfair, let alone the standards of 3 years ago.
We now have professional mathematicians that are on the record carelessly dismissing major math questions being decided. And I'm not talking about having a subtle discussion about what proofs mean.
So it's not really about trying it out, I think. It's something more emotional.
The reality is that even if they used the tools, they still wouldn't change their mind.
Many people have incredible ability to ignore evidence.
I'll just point at this essay from five months ago and hope you folks read it carefully. [0]
I'll include this excerpt from the essay as an attempt to shame folks who feel the urge to continue to claim that the author is critical of "AI" because they hate the tech and never ever have anything good to say about it:
> I will say that Claude Code is a noticeable step change over the Opencode-based GLM models that I tried a few months ago. Those were bad enough to not be usable for anything serious or of real complexity. Claude Code, by contrast, can do some relatively complex things: it was able to perform the bulk of my website migration from Nuxt 3 to Nuxt 4 and to the newest version of Nuxt-Content for instance, which the GLM models had utterly barfed on. It wasn't perfect: I had to do quite a bit of tinkering after the fact to make things work, but unlike the GLM attempt, Claude did actually save me some time. Claude also did pretty well with adding Schema.org markup to my website and performing a number of other stupid-but-necessary SEO-related tasks on the same website: none of these tasks are really groundbreaking, but they're the kind of relatively complex pain-in-the-ass task that you'd hope we could automate: at this point in my practice, there's little of value that I can glean from doing those tasks by hand. If it's possible to make this kind of tool accessible to people at an acceptable, sustainable cost and without burning too many resources, then, I think that'd be a good thing.
> Claude Code also does quite a bit better than Opencode tools did at CI/CD tasks, IaC in OpenTofu, building Dockerfiles and that kind of task. It's still something I'm quite iffy on, but unlike the GLM models I tried earlier, what Claude Code produces basically works, at least initially. This probably makes it a lot more immediately useful than the GLM models were, as this kind of task is a large part of what tech workers actually do day-to-day, and it's the kind of task that a software engineer might have to perform while not having a very good grasp of the underlying technology.
I recommend folks take 15->30 minutes to actually read the essay, rather than either skimming it or having some LLM "summarize" it. There's a lot of subtlety and nuance that the current generation of lossy compressors is incapable of reliably compacting.
[0] <https://deadsimpletech.com/blog/engineering-judgement-claude...>
> The technology can, in volume of adequate output or in quality of output, overwhelmingly outclass what alternative technologies can do. A moderate improvement isn't really enough to force inevitability: it has to be a massive difference.
> Now we come to the question of whether or not the difference that LLMs bring to the table is sufficiently overwhelming to make their adoption inevitable, and on this I think the answer is a clear no. As established above, LLMs just can't do top-flight work at all: at the heights of basically any discipline, we're still very much relying on unaugmented human work.
The clear flaw here is that cost isn't taken into account. A new technology needs to be better than the status quo at a given cost/price point.
Examples: mass produced textiles, photography, IKEA. Ostensibly not the highest quality, but the cheapest at a price point.
This was published today, I had to check the date to be sure. I used to be an applied researcher working in big tech and for the things I wanted to build if I had to pick between LLMs and hand picking a team of 100 from the best programmers / researchers in the world I would pick the LLMs. Of course the combination would be even better. I've had LLMs immediately understand concepts that took me months to effectively communicate to a colleague and there are some concepts I was unable to communicate to any other human at all. Then the LLMs could build what I needed in less time it would take me to set up a meeting about it.
> Any actor that's at all serious in their field is, sooner or later, going to hit a point where LLMs cease being useful to them and actively hinder their further progress, and we absolutely do not see the massively overwhelming advantages that LLMs would need to have in order to become inevitable.
I can only conclude from this that I am not at all serious in my field. That's okay. I'm too busy enjoying myself.
Ironically, this might be the first time I wished an article was written by AI.
[flagged]
If you don’t use AI to scan for security bugs yourself, someone else will likely be using AI to find them first. So in that way, at least, LLMs could be considered inevitable.
I know some colleagues that refuse to use AI, even for security review. They're piling up security issues, commit after commit.
The first part of the article is a thoughtful framework. The conclusion that LLMs are not inevitable because they aren't useful or that their utility can be substituted by humans is an extraordinarily flimsy claim.
Human intellectual effort is the tightrope walker in Thus Spoke Zarathustra; AI is the jester. LLMs can already solve programming and mathematical problems better and faster than humans. It's only a matter of time before other intellectual disciplines fall. We would be profoundly stupid not to embrace this change.
Empathy and big-picture thinking over myopic specialization will be the new moat.
I read "The level of social opprobrium directed at LLM use and at people attempting to pass off generated slop as work is immense" and immediately checked for the author's Bluesky link. This person is living in a bubble. Most people in tech are now using LLMs at work and there is no social opprobrium discouraging them.
The rest of the analysis seems to take it for granted that we have hit a wall and LLMs will improve no further. OK. We'll see how that prediction plays out.
> Most people in tech are now using LLMs at work and there is no social opprobrium discouraging them.
I don't know if "most people" is true, yet. I do think that the opprobrium has been silenced, and in corporate environments that has been systematic and top-down. Tokenmaxxing, leaderboards, monitoring: this is all part of making sure developers do not spend time considering the ethics, because the conclusions have been rendered moot.
Something like 90% of devs are using LLMs. It's how software gets written now, end of story.
> Something like 90% of devs are using LLMs.
That needs a citation, and context. Globally? The USA? The Valley? Where is this number from?
JetBrains developer survey 2026:
https://blog.jetbrains.com/research/2026/08/ai-coding-agent-...
"90% of professional developers were using AI coding agents at work at least weekly in one form or another (local agents or remote cloud agents), with 68% using them daily."
Re. context:
"The Developer Ecosystem and AI Pulse surveys are localized into eight languages: English, Spanish, Chinese, Japanese, Korean, German, French, and Portuguese. We apply quotas on the required number of responses by region to help achieve accurate global representation. ... The Developer Ecosystem Survey has been statistically reweighted to better represent the global developer population by region, employment status, programming language, and familiarity with JetBrains products"
Fully unclear who, actually, is being surveyed. Developers generally through some industry-wide survey? Or JetBrains users?
(The survey was run on Jetbrains instagram page, apparently. I mean… that is a bit of selection and narrowing going on there)
Notice how half of them have heard about Jetbrains!? I don't know any developers I would be confident know about Jetbrains (even among AI-using developers)
(Also the graphs don't really support the idea that 90% are using them meaningfully for software development.)
We need to start talking about anti-AI psychosis. That people actually think their delusions fueled by these echo chambers are real is almost as worrying as the AI booster bubbles.
I'm gonna go ahead and make the prediction that this eventually morphs into a kind of religion, like flat earthers.
Nothing that AI will ever be valuable, everything AI does is steal, nothing it creates is new, everything that the big labs publishes is a scam or hype. Oh and "it's just doing X", where you can substitute "X" with some kind of surface level description of how the output layer in an LLM works.
These tenets are already pretty much formed, all it needs is to be formalized by some kind of name or manifesto similar to Q-anon, and we've arrived at a durable state of permanent denial.
There‘s hardly anything to talk about, since most of those anti-AI people use LLMs anyway. It‘s hypocrisy at best. If only that energy could be channeled in doing more about open-weight models-we need more AI-ownership, not less AI. That LLMs will somehow magically disappear should Anthropic or OpenAI go bust, is nothing but wishful thinking.
They need a holy text.
Maybe an Orange Catholic Bible.
One thing I've seen with LLMs that I've not seen with any other emergent technology is the level of backlash against it. I think we're seeing a point of polarization in society where there will be two truly opposing sides that are against each other and no reconciliation will be possible between them (pro and anti-AI), and that it has a good chance of fundamentally destabilizing society.
> I've not seen with any other emergent technology is the level of backlash against it.
i think those who had backlashes against machines and automation in the past would be similar tbh.
It's because when people feel, even if remotely, their livelihood going down the drain, they do anything and everything to fight it (even if it would be against the interest of overall progress - nobody wants to be the "sacrifice").
I mean, Luddites were hanged for murdering people. We're not quite there yet.
They were not just fighting for their livelihoods, though; the fight was at least as much about dignity and fair treatment.
And they were correct to do so: skilled workers who lost their work and ended up in industrial employment did not have better lives. It took a century to a century-and-a-half for industrial quality of life to get back to pre-industrial quality of life.
Loss of dignity is something that the AI industry is not considering as it renders entire swathes of the economy button-pushers for chatbots.
When loss of livelihood appears it will be too late and at too great a scale to fight back in the workplace; the fightback will be somewhere else.
H.G Wells Time Machine has future society split into to: the childish consumer class and the driven engineering class, and the consumer class is the food of the engineering class. I could see this happening, with how cruel so many are, and trying to institutionalize that attitude.
Silicon Valley tech workers are not the subterranean Munrungs. They are the surface-dwelling Eloi. They just don't know it yet.
I've seen a lot of people not wanting to learn to use a computer ("I prefer the old way") or not wanting to have a mobile phone ("I don't want to be reachable all the time") or a smartphone with touchscreen ("This feels like a too close interaction between the human and the device").
> One thing I've seen with LLMs that I've not seen with any other emergent technology is the level of backlash against it.
Blockchain and other cryptocurrency related technologies?
I'm pretty sure that an average person has no idea what blockchain is.
The average person has no idea what LLMs are. The average person (in the west at least) has heard of and has an opinion on ChatGPT and Bitcoin.
It's not a normal technology.
I haven't seen much splitting into two groups of people who are for and against. It's more that individuals can see upsides - does cool stuff, and downsides - something like a rival lifeform that will be smarter than us.
I think that's a pretty odd framing, because the upsides are pretty relative and lopsided to those receiving certain benefits. The people who are anti-AI often are those who can acknowledge that AI does cool stuff, but that it's not worth it and doesn't provide any real benefit except by helping people keep up in a frenetic arms race to develop new tech.
Personally, I'm against AI but even if it could never be smarter than us and if it were at the state that it was last year, I'd still be against it due to the environmental cost.
I'm mildly pro but agree on the environmental costs. I think they should try to make it green energy only though not much chance with the current US admin. I'm not so enthused with current AI. I use last years like language translation a lot and I'm interested in the possible sci-fi future like curing diseases, some sort of immortality, abundance etc.
I keep seeing it and it's really bipolar. Boosters think it's the best thing ever and downplays all the issues while antis think it's completely useless and a disaster environmentally and morally.
Good luck saying LLMs are useful but still a net negative in the filter bubble this author is in. No, you also need to think it has zero utility or you're the enemy.
One datapoint on the same people seeing pro and anti is Nick Bostrom (Swedish philosopher & writer) who was one of the instigators of AI doom with his 2014 book:
Superintelligence 2014 https://en.wikipedia.org/wiki/Superintelligence:_Paths,_Dang...
has now brought out another book on the positives:
Deep Utopia 2024 https://ndpr.nd.edu/reviews/deep-utopia-life-and-meaning-in-...
I was just watching an interview with him:
Nick Bostrom Says We Are Clueless About What's Coming https://youtu.be/J9WBZ8BTLLg https://www.nytimes.com/2026/10/01/opinion/interesting-times... - rattles on a bit
Apparently Yudkowsky was influenced by the 2014 book in coming up with his stuff.
You missed Gnome v KDE
Well yeah, that's because it's political. The theat of job (and more importantly income) losses as well as twitter accelerationists, AI art slop spammers and anxiety trolls have really fueled the fire to the degree that people hate this stuff.
Not to mention how creepy Sam Altman and Dario Amodei are.
The author's absurd fantasy of how an LLM has to be used in order to fit their imagined adversary's idea of inevitable is absolute drivel. This article doesn't even build a straw man. It's just a pile of straw and manure.
“That surveillance capitalism is a logic in action and not a technology is a vital point because surveillance capitalists want us to think that their practices are inevitable expressions of the technologies they employ. For example, in 2009 the public first became aware that Google maintains our search histories indefinitely: data that are available as raw-material supplies are also available to intelligence and law-enforcement agencies. When questioned about these practices, the corporation’s former CEO Eric Schmidt mused, ‘The reality is that search engines including Google do retain this information for some time.’ In truth, search engines do not retain, but surveillance capitalism does. Schmidt’s statement is a classic of misdirection that bewilders the public by conflating commercial imperatives and technological necessity.”
Shoshanna Zuboff, “The Age of Surveillance Capitalism”
> In papers filed with the court on November 20, the DOJ told the court that Windows 95 would function without Internet Explorer 3.0, and that Microsoft should be required to offer computer manufacturers (OEMs) the option of installing Windows, free of Internet Explorer 3.0 in “all” respects. In response, Microsoft explained that the latest versions of Windows 95 would not work under such a scenario
https://news.microsoft.com/source/1997/12/23/dojs-request-fo...
Cope.
[flagged]
It helps if the author does not confuse what LLMs actually do with the general misunderstanding that "LLMs do work". They do not "do work" and have never "done work", and this misunderstanding is all over tech.
What do LLMs to? You know this: they are sentence completion engines, that through this process of sentence completion can be used to create software than then "does work". This indirection is critical, it is "how this works" and is the true power of LLMs that is STILL not being recognized. The power of LLMs is their communications capacity, their ability to complete sentences, which is so subtle the entire world of software engineers and artificial intelligence everybody has not identified that simple "variable expansion" as the true power of LLMs. LLMs do not do work, they enable the creation of software that then does work.
With a technology so subtle, want to know what is inevitable? A bifurcation of the population into those that believe in magic and those that understand it well enough to make magic nobody considered possible. Why did I say "magic"? Because people are giving up understanding, and that makes what we do magic.