Ahh yes, the "human dignity" of billions of people toiling away to make Americans cheap widgets. There never was dignity to capitalism. We can strive to replace it with something better.
If you paid attention to proponents of a wealth tax in the US, you would be aware that they only ever suggest it for vast wealths of like $10 million+.
The Bay Area is one of the most expensive parts of the United States and $10M still means you own half a dozen houses. I’d say that’s reasonably “wealthy” from the perspective of the majority of the population who struggle to afford even one.
Anyone who owns two $500k houses is wealthy in 2026. I used to own two worth less than that and didn’t consider myself wealthy, but I was by the statistics.
Yeah the multi-agent workflow just hasn't been satisfying to me. The more chats I try to run at once, the more I got lost and overwhelmed. I trust Claude to implement a plan correctly after I've reviewed it, but if I don't review all of the plans, I will miss some small detail that it misunderstood and it'll be a pain to fix later.
I'm like a 1-2 chats at a time kind of guy. I just don't see how I could keep my exact vision for the project otherwise.
Same, on top of that multi-agent workflows just cost too much to make stopping and correcting them to feel worthwhile, compared to one or two manually managed chats
Counterpoint: Work sucks. Of the billions of workers on the planet, the number of them who love their job and would truly be doing it even if they didn't need to in order to survive is probably in the low single digits.
Hating work is good, wanting it to all be automated is good. It is a pro-human flourishing stance, whereas keeping the majority of humanity laboring in jobs they dislike just to survive is against human flourishing in favor of the status quo.
I don't think many people disagree with this. The main problem is that labour has been what allows regular people to have negociating power with those who own most of the capital.
People are worried that if they lose this leverage, nothing is stopping the few who have most of the capital to just disregard the needs of the masses.
Democracy is what allows regular people to have negotiating power vs the rich, and the majority of these battles are actually won through legislation, not union negotiation.
I understand that regular people have lost faith in democracy, and that they think rich people control the world and make every major decision, but that just doesn't ring true to me. Democracy is more or less giving us what we vote for, we just vote for dumb things. Ultimately, I have faith that if political and economic circumstances change enough, we might actually vote for the right things.
> Democracy is more or less giving us what we vote for, we just vote for dumb things.
Sorry, when did I vote in favor of the Citizens United, McCutcheon, or Buckley outcomes, all of which are tied to our current predicament of money in politics? And how is my voice, through my vote, able to change and possibly overrule those decisions? Until that happens, even the ballot box won't give me negotiating power versus the wealthy.
The five Supreme Court Justices who voted in favor of Citizens United and McCutcheon were all appointed by Ronald Reagan and George H.W. Bush. Had either of those two not won the presidency, Citizens United and McCutcheon would have lost their cases against the FEC.
People can still vote for representatives who want to pass legislation to undo those cases through congress. There's nothing that's stopping people from voting in representatives who are in favor of this (yet).
When the mine closes, democracy does not save the town. It still becomes a ghost town. Democracy cannot work upstream against fundamental market forces.
Every industry is being pitched as becoming a ghost town. Democracy will not save us. We will all die in the future and not be replaced. This is the end of our line in this century. I hope you are proud of how far we’ve come because this is the end for us. There will be no economic justification for humanity quite soon and we will likely be slaughtered to eliminate latent variability.
That's not a very good analogy. When the mine closes, democracy doesn't save the town, because the town only has control over the economic activity inside of it.
AI or no AI, the US will still have control over the economic activity inside of it. America is not pitched to be a ghost town, but instead a vastly productive and wealthy economy driven by AI. A ghost town cannot redistribute wealth it does not have, but America will still have all of the wealth.
Democracy is also doomed by sufficiently capable AI. When the "meta" military unit was a knight in shining armor, most societies were under feudalism, ie rule by knights. When guns became cheap enough that whoever had the most guys would win a civil war, we got democracy: rule by whoever has the most guys. When whoever has the most robots will win a civil war, what kind of government do you expect?
It's impossible to predict when they happen, or their outcomes. The world may be worse at least for a while after them. Or they fail in general.
But they happen and then all the people who were crowing about the inevitability of some existing order and now it embodies natural law and what not look really f*ckin stupid in retrospect.
People believed in the divine right of kings with the same full earnestness of people on this forum who have think AI is just the outcropping of some transcendent mathematical telos.
I might disagree with it, unexpectedly, even though I'm very lazy and anti-work and would have agreed with it ten years ago. This isn't some they took our jobs stance, either.
Thing is, you have this mythical beast, the "dark factory". This exists mainly as way to humiliate the west by suggesting that China is way more developed. One reason that it's unlikely to be substantially real is because of the failure of robotics to really replicate adaptable, self-repairing, sensitive, sensible humans in an industrial context. But two of those adjectives are technical, while the other two, adaptable and sensible, are to do with knowledge and creativity.
I mean that it's an ugly fact that human creativity (thinking on your feet), and morality even (knowing what to do), is useful and necessary in the context of the most boring shitwork. Even on an assembly line, if you're expected to do some QA and accept ad-hoc instructions for different products. I don't want us to be diminished by having to do the shitwork, but I don't think AI can make it go away.
Oh come on, why a downvote? I put some thought into this and all I get is a binary nah.
You're directionally right, anyway. There's no reason significantly advanced AI (likely to not be developed from LLMs but from some other path) can't completely replace a wide variety of human labor. But replacing human labor with machinery (i.e., capital) is not new, it's been going on for a couple centuries plus some. The thing that happens when you replace wage labor with capital is that the rate of profit (i.e., the ratio of profit to the amount of invested capital) tends to fall, which is a systemic threat to investment. The recurring tech and asset bubbles since the 1990s have each been inflated in an attempt to maintain rising levels of investment in the face of rising productivity and therefore falling profit. An economy of dark factories isn't useful under capitalism, because it produces goods which end up having no sale-value.
The Musk-esque theory is that robots can do all the drudgery for free, while the economy centers on humans being creative, which in this trope looks like an Elysium with singing and dancing and poetry and painting, and maybe togas.
The core argument is that people don't want to be machines, and shouldn't do mechanical work, and that it's a shame if anybody feels compelled to work like a machine to survive. But then we have the job loss part, in which, because of automation, that person doesn't even have the option of working like a machine, and complains about it.
However, I'm coming round to thinking that the vision of a continual symposium-party is wrong anyway. I think automation can't do all that much, and creativity is needed in even the most mundane and dreary contexts. This means automation is less disruptive that it's purported to be - but is still somewhat disruptive - and the change in the nature of jobs is less of a shift to creativity - but is still a small shift to creativity. The jobs aren't delightful, people are still needed in factories, and there are no togas involved. This is my dreary insight.
I often hear people talk online about burning data centers to avoid some capitalist dystopia.
It just seems incredibly pessimistic to me. Who wants civil unrest? The rich elite does not want this either.
We will pay people.
Capitalism is not set in stone when human labor is no longer essential for productivity and AI can handle planning that markets currently coordinate through capitalism.
Exactly! The rich don't want to see mass starvation any more than the rest of us. We only permit homelessness and food insecurity now because of scarcity and a "just deserts" mentality where we blame people for their lot in life. When AI is doing the majority of labor, there will be no "just deserts" mentality, and there will be massive abundance.
This is plainly delusional. There already is abundance, global crop lands produce enough calories to feed twice the world’s population[0]. Greed is the reason for inequality and “AI” is not solving that. It is pure wishful utopian thinking to believe that there will be some massive AI-initiated abundance.
Speaking specifically about food insecurity and homelessness in the US, it's not simply greed, it's "just deserts" ideology. It's a belief in the lack of merit of the poor to receive help.
Speaking globally, there are many more barriers to feeding everybody than just abundance, like the other guy said.
> Hating work is good, wanting it to all be automated is good.
Not without a concrete answer for how we all continue to survive and thrive when our jobs are replaced. And that's the part the AI boosters are silent on, beyond vague notions of UBI.
I’m not sure work sucks (at least as a general statement). Work is a way to be useful in the society and feeling useful is valuable for our wellbeing. I don’t think people who stop working and spend their days laying on the beach feel better.
> Hating work is good, wanting it to all be automated is good.
It's this "path of least resistance life style" that ruins a lot in our society. Yes, being wasteful is bad, but not appreciating the work is also bad.
While that may be true, all the automation that's happening is not making people work less.
Except by firing them, which is absolutely devastating, especially in places with 0 social protection like the USA.
And for other people the workload has actually increased because we're being pushed left and right to use AI to go FASTER, yet over and over we see proof that we need to validate EVERYTHING.
Automation is only useful if we use part of our brain to do something ELSE. Yet as a Software Engineer I cannot do that (full disclosure: I don't mind, I like to engineer and code).
But from a managerial point of view there is this aura of 'AI will solve all our problems', whereas what we're seeing is a deluge of slop where the people producing the slop THINK they're helping, yet they are parasitizing on the time of the people who validate the PRs.
Yeah but the tech doesn't end here. Sure, it doesn't make anybody particularly happy that the tech started here. It's kind of a bummer to automate creative labor that gets social esteem so early on.
But that's not the point. Where the tech ends is human-level intelligence that can do any task as well as a human and more cheaply.
It's highly cynical, but people need to work. It provides structure and most people don't do well with unstructured time.
Also, I find it odd that of all the automation being attempted with LLMs, we're automating the ones that actually are interesting, not the ones that are dangerous or truly rote, yet highly mechanical.
I would humor this stance if we were also actively building a new economic arrangement that was not capitalism.
Automating away the drudgery or dangerous parts of life seems inherently good. But I would argue that AI has not been awesome at that, really. There are certainly cases where it has lessened tiresome work, but there are just as many cases where AI is worsening the pleasant parts of life. And I don't know anyone who has experienced shorter work weeks because AI is doing stuff for them.
Under capitalism, AI is converting labor power of ordinary people to "property" owned by the owning class. It is making the rich richer. It doesn't really improve my state of being.
Humans will not flourish if you remove their jobs, they will become violent criminals because they will have nothing else to fill their days and no purpose in their community.
People may hate their job, but they will hate being unemployed way more.
People can find purpose without jobs. But they can't find purpose if they are struggling to survive. If jobs are the only legal path to survival, and there are no jobs, then people will be driven to "crime" to survive.
It's an informed view for anyone who's spent time around multi-generational welfare-dependent households. Regardless of race or creed, the majority descend into substance abuse and domestic violence.
Exactly, and that's why this maxim about "understanding the code base" being the bottleneck is also somewhat misleading.
Claude is even better at helping you understand the code base then it is at writing code! It can look at a bunch of files and give you an accurate run down in ten minutes.
We don't need AI in the same way we don't need washing machines and dryers. Like, sure, we don't need a machine to do our laundry, just like we don't need an AI to do our skilled labor, but it sure saves us a lot of time and energy.
All subscription models are subsidized by users who don't use much. The fact that somebody on a $20 sub might get $50 in value isn't crazy if there are 3 people who only get $10 in value. This isn't some sign that the model is broken, it's the intended outcome.
Also, I didn't read this whole thing, but I have yet to see Zitron respond to the strongest AI financials claim, which is that the models themselves are profitable on a life-cycle basis, even if the companies are not profitable on an annual basis due to capital expenditure. Dario made this claim exactly, and it more or less blows all of Zitron's financials arguments up.
Thanks for the link. I'll admit I'm not an expert on the business side of this, but is this really much of a response? He seems to just call it strange accounting and then he moves on.
It doesn't even feel like particularly strange accounting to me. Aren't there plenty of companies that spend a lot in one year and realize the gains in the next year? If I build a house this year and sell it next year, the house was still profitable, even if next year I'm building 3 more houses to sell in the year after.
The TL;DR is that Dario likes to talk about imaginary/hypothetical companies a lot in interviews, and those companies' financials don't have a direct basis in reality.
Thanks for the link. There's not much of an argument here from Ed, though, besides that it's an unusual way to view or report margins.
But it's not that unusual, right? If I build a house this year and sell it next year, the house might still be profitable even if next year I'm building 3 more houses, so the company as a whole is still in the red on an annual basis.
I mean, I'm not a financial expert but that doesn't seem all that unusual to me.
The first part of the argument is just noticing that Dario is carefully avoiding making factual claims about Anthropic. Like, if the bank asked you if your construction company was profitable, would it be acceptable to respond: "Well, hypothetically, if a construction company sold houses for more than it cost to build them, that company could be considered profitable. It is possible to imagine a stylized model of a construction company that is theoretically profitable."? If the real, non-hypothetical company that Dario runs has financial results which support this argument, he should probably say them more often.
The second prong of the argument is basically that, when you invest in Anthropic, you can't just invest in one model and then collect the profits from that model. You're investing in a whole company in the hopes that they can be profitable overall; at some point they'll need to stop spending so much money on training and give it back to the investors instead. Zitron argues that this isn't going to happen because training is actually something that companies need to do to retain customers at all. An analogy here might be the fact that Microsoft has to spend a certain amount of "R&D" budget fixing security vulnerabilities in Windows Server just to retain their current customer base; if attackers found out about a serious security hole but Microsoft didn't fix it, everyone would need to stop using Windows Server. LLM companies do the same kind of thing to fix "jailbreaks" and other unexpected model behaviors.
The third prong of the argument is that, in general, there's a long history of companies using creative accounting to try and make themselves look profitable and then collapsing because they're not actually profitable. For example, WeWork's "community-adjusted EBIDTA" figured claimed the company was profitable using very similar arguments to Dario, and then the company went bankrupt. If you're already cooking the numbers, you have almost arbitrary flexibility to report whatever "margins" you want by excluding some of your costs from the calculation.
> hypothetically, if a construction company sold houses for more than it cost to build them, that company could be considered profitable.
Construction companies capitalize and depreciate over many years so they can answer "yes" they are profitable even when they are very cashflow negative. This is exactly Dario's point: model training costs are treated as expenses but in practice are much closer to construction costs. Model training effectively produces an asset, the model weights, which will generate revenue for many years into the future.
> Zitron argues that this isn't going to happen because training is actually something that companies need to do to retain customers at all.
This is exactly why Dario's point about each training run being profitable is so important. It suggest that this is not true. Customers are happy to use old models long enough to fully pay off their costs.
> there's a long history of companies using creative accounting
Zitron seems to know very little about accounting evidenced by him using terms like "gross margin" wrong in this article. He's pattern matching against his limited exposure to company financials to find superficial similarities between the AI labs and famous frauds. Find me a company that doesn't report non-GAAP measures. Google search claims 96% of SP 500 companies do it. Are they all frauds too? Sometimes non-GAAP adjustments are eye roll inducing but they are tolerated because they can be genuinely useful to get a fuller picture of the business.
Thanks for the genuine response. When you put it like that, though, if all seems a little ambiguous. Like, Dario isn't necessarily lying, and there's no proof he is, and on the contrary, the company continues to get investment from people who, in theory, do get to see the actual numbers.
I guess I don't blame you or anybody for having a deal of cynicism, but these arguments just don't seem very concrete. Like, if Dario was lying or not, he probably wouldn't share the actual numbers, and he probably would propose a "model lifecycle" accounting. And if the business model had potential or not, there probably would still be vast investment in the next model. Zitron has had nothing but cynicism towards AI from the start, and it's his whole shtick, and so these arguments don't seem very credible coming from him, even though they seem reasonable coming from you.
FWIW I personally think the most reasonable take is basically that the things Dario says should not move the needle on whether or not you believe Anthropic is profitable. The things he is saying are indistinguishable from the things he'd say if Anthropic was not profitable.
I subscribed to Claude for a month. I sat down with it for a few sessions, but in each case I ran into a limit before I achieved anything worthwhile. And that was with me babysitting it the whole time to try to get the most out of it. I'm not sure it's possible to use it less (so that others can use it more) and get anything meaningful done.
Most small features take 80-150k tokens to implement, and most large features take 200-250k. For a hobbiest working like 10 hours a week, they can get stuff done but not nearly hit the weekly usage cap.
> which is that the models themselves are profitable on a life-cycle basis, even if the companies are not profitable on an annual basis due to capital expenditure.
Until they file an S1 to go public and show the world the books, take everything they say with a grain of salt. The amount of financial engineering going on in this space is astounding, and I'll believe it when I see an objective 3rd party release an audit confirming this claim.
Their bigger incentive is to deliver the best product in the cheapest way, because there is tight competition with at least 2 other companies. I know we all love to hate on capitalism but it's actually functioning fine in this situation, and the token inflation is their attempt to provide a better product, not a worse one.
The whole idea of just sending "no" to an LLM without additional context is kind of silly. It's smart enough to know that if you just didn't want it to proceed, you would just not respond to it.
The fact that you responded to it tells it that it should do something, and so it looks for additional context (for the build mode change) to decide what to do.
I agree the idea of just sending "no" to an LLM without any task for it to do is silly. It doesn't need to know that I don't want it to implement it, it's not waiting for an answer.
It's not smart enough to know you would just not respond to it, not even close. It's been trained to do tasks in response to prompts, not to just be like "k, cool", which is probably the cause of this (egregious) error.
I didn't mean to imply that it was. But when you reply to it, if you just say "no" then it's aware that you could've just not responded, and that normally you would never respond to it unless you were asking for something more.
It just doesn't make any sense to respond no in this situation, and so it confuses the LLM and so it looks for more context.
No, it has knowledge of what it is and how it is used.
I'm guessing you and the other guy are taking issue with the words "aware of" when I'm just saying it has knowledge of these things. Awareness doesn't have to imply a continual conscious state.
You know why they don't share the fruits of capital with us now? Because Americans hate getting taxed to pay for welfare, and so they've been voting against taxes for 50 years. This whole political landscape changes when people lose their jobs to AI, a thing that everyone thinks should be taxed. In fact, the entire ideological underpinning behind extreme wealth accumulation is gone when AI runs everything.
Works great in other countries with high unemployment. That's exactly what happens! People elect a person who says they are going to change everything to fix it and they never get around to it for some reason :)
Yes, because current unemployment comes as the result of complex factors that intersect with various groups and ideologies in complex ways. Also, raising employment is a complex task.
When AI takes the jobs, it will be dead simple to the majority of people that the current way of doing things will not work, in the same way it's dead simple to you and me.
reply