I'm not usually for arguments of "this money could have been better spent elsewhere", but here's a thought experiment. Lets say instead of injecting $2 trillion and counting into a few AI companies, we instead injected $2 trillion dollars into things like infrastructure (real infrastructure, not GPU warehouses), education, helping out communities ravaged by globalization (I doubt most people on Hacker News venture outside of coastal areas, but if you want to make a REAL difference as an entrepreneur why not look at parts of the country that are struggling and figure out how you could make a difference there? You know, instead of trying to just ruin the economy for the sake of the already-obscenely-wealthy). I'm not saying all those ventures would succeed, but I think that amount of money put towards boring-but-real problems would make a much bigger positive impact for everyone.
This is 100 manhattan projects. Lots and lots and lots of things that can be done. Reminds me of the quote "water water everywhere not a drop to drink" ==> Money, money everywhere, not a cent goes towards something remotely useful.
1) Could've added nifty features to roads (rf beacons?) and make it super easy to build autonomous vehicles. This would've solved so many problems, but most importantly 45,000 lives every year and millions of accidents.
2) Could've given the money to farmers who are getting ethanol subsidies today to add panels, supply tons of power to the grid, creating permanent income to the farmers instead of handouts every year.
3) Could've built many Universities permanently free which is feasible with an endowment that large.
4) Could've built walking/biking infrastructure in most cities, which would help reduce healthcare expenses. Most of the expenses today are lifestyle/metabolic diseases.
5) Could've switched the entire country to clean energy, exporting ALL of the oil and create a sovereign wealth fund.
6) Could've built a city for homeless people with homes, excess energy production (solar) providing the function of a UBI, a massive hospital with health and mental health facilities, de-addiction centers. New city in the middle of nowhere where land is dirt cheap. Army corps of engineers could execute this.
You are describing altruistic gifts, not investment meant to generate a return.
All this money is being deployed to try and turn a dime into a dollar.
The biggest area I can see where money could be deployed into building something useful -- while generating a positive return -- is real estate development in cities with massive housing shortages. That's basically not allowed in the same cities. In large part, because people who yearn for an more altruistic world care very much about their own quality of life and property values than they do about the well-being of their community in the long run.
> For every £1 invested, walking and cycling return an average of around £5-6. A ‘benefit to cost ratio’ (BCR) greater than or equal to 4 is considered to be ‘very high’ value for money. Putting this in some context, the predicted BCR for the HS2 (high speed railway) wasn’t nearly so impressive at about £2.30 for every £1 invested.
It's private money. At least in my country the government isn't giving a single cent to AI outside of fundamental research projects.
If companies want to bet all their money at roulette they are free to do so.
You have to wait 20 years for the returns to society. Public education was enormously successful when it was introduced in the 19th century. There's just no profit in waiting for second order effects to kick in.
What makes you choose the 19th century for public education creation? It was my understanding the Puritans and thusly Yankees that spread through NE and the north Midwest were staunch proponents of schoolhouses and publicly funded education dating back to the to the early 17th century.
So was Rural Free Delivery. Farmers being able to communicate was a massive boon. There is a channel for farmers called RFD tv. They completely scrubbed the free provided by the government part after private equity bought the tv channel targeting farmers. Then they got Imus in the Morning so farmers listed to Imus, Rush, Hannity, and orielly forgetting the government helps them.
Diminishing returns. Per-student education spending has been going up since 1990 except a dip during the 2008 recession. Adjusting for inflations it’s now double what it was 30 years ago.
Why would they be any better than the people distributing the money to them in the first place? The students working hard at "providing the good educational outcomes", by this economy's standards, are suckers.
Investment in education is one of the best ROI a society can have. Quite surprising that in the USA this relationship has been eroded by schools spending in sports and admin, in most other developed countries it's a quite direct relationship between more investments -> better educational outcomes.
But there's 2 problems with that. To save money governments declare that "teaching is a profession". By which they mean: "teaching", not subject matter expertise, matters. Except:
1) the single thing you cannot beat is capable teachers. And NOT capable as in able to teach, subject-matter-experts. And that is what everybody wants to avoid at all costs.
Put it like this: an unwashed, abrasive, offensive smelly asshole with a master in math can teach kids math. A super-friendly, nice, beautiful, polite social science major (or not even that) who doesn't know math well cannot. The level of teachers needs to be an order of magnitude above the level expected of kids at the end of the class. Knowledge of the subject needs to take absolute precedence in teacher hiring decisions, to the point that good teachers get rejected if lacking expertise in the subject.
This is not how things work. They only worked that way for a short while due to the crash of 89. I've had a French teacher that couldn't speak more than "some" French. Certainly nowhere near fluent.
2) Money needs to support that. Which means for Math, the money for math teachers needs to be competitive with research departments of banks, and very different for different subjects, which seems offensive to schools and government in general. If teaching positions aren't competing with other professional careers in the subject matter, it will never work. This will, of course, mean that many science teachers make more than a school principal and even a district administrator. Deal with it.
For Math, they are literally not even 25% of that, but even for French it's insufficient.
> 2) Money needs to support that. Which means for Math, the money for math teachers needs to be competitive with research departments of banks, and very different for different subjects, which seems offensive to schools and government in general. If teaching positions aren't competing with other professional careers in the subject matter, it will never work. This will, of course, mean that many science teachers make more than a school principal and even a district administrator. Deal with it.
I believe this to be a heavily US-centric view, Estonia and Finland do not pay more for math teachers than they could get in banks' research departments but still manage to get competent teaching across most school subjects.
Things work outside of the US, the issue is the US does not want to look outside and adapt itself to incorporate what's been learnt from other systems. It's a "do the US way or no way" attitude, and it's slowly eroding to become more a "no way" than any kind of "US way" of doing things.
Doesn't change the fact that I have 2 daughters that are getting Math education from someone who is easily beat in math teaching quality by Google Gemini (I realize this is a specific strong point of Gemini, but still)
Well yes, you have to spend the money wisely. How could we construct a system so that we have 2x as many teachers (thereby halving the classroom size)? That would have a lot of good second-order effects beyond test scores.
So why has per-capita student spending doubled since 1990 (adjusted for inflation) without any increase in test scores? Why haven’t we been spending the money wisely?
> we instead injected $2 trillion dollars into things like infrastructure (real infrastructure, not GPU warehouses), education, helping out communities ravaged by globalization
Even excluding military spending, US governments spend $2 trillion every 10 weeks.
I wasn't making a judgment about including or not including state spending. It's just that "US governments" is not a common way for Americans to describe federal and state. People think federal when they see "US government".
You have to include state and local spending too. We’re at 40% of GDP which works out to almost $13 trillion: https://fred.stlouisfed.org/graph/?g=1CFpQ. Subtract $1 trillion in defense, and we’re spending about $1 trillion a month on government.
Does that chart double-count state transfers to municipalities? When I was in local government, about half our school budget came from the state, so there would be entries on both ledgers.
AI datacenters don't really confer any geographical benefit since their output can be transmitted at extremely low cost. The benefit of local datacenters is purely political and has nothing to do with technological advantage.
People invest in what would get them the highest ROI. No one is really thinking about “improving the world”* when they invest.
What’s scary about the AI boom is people over investing and not being able to recoup their investment which will lead to knock on effects - companies going bankrupt and people losing jobs, savings gone, … etc.
* As ESG has shown, not everyone agrees on what is considered “improving”.
It won't make much difference. The US has a lot of problem, but "not spending enough money" isn't one.
The US government spend a lot on healthcare ($5.3 in 2024)[0]. More than most European countries per capita. But many people still feel that the US hardly has healthcare at all. Pouring more money without a full structural overhaul will likely make things worse.
And the $2T you mentioned is investors' money, which means that your plan is actually to increase tax by $2T and pour it into a system proven inefficient.
It all sounds great, but unfortunately human nature is that money attracts corruption, so how could a $2T injection be managed in a way that ensures everything is square, without adding significant overhead to the spending?
Governments are often just as bad at this as private entities are.
It's quite easy. The money you "inject" needs to have demurrage.
You can then give all of the $2 trillion demurrage dollars directly to the most corrupt individual on the planet and it won't matter.
There are two scenarios:
1. The recipient keeps the money: It will automatically be returned.
2. The recipient spends the money: It has flown from the most corrupt individual on the planet to a less corrupt individual.
If we assume that each transaction allows the next recipient to make a decision on how to best spend the money, it would automatically spread out, since it cannot stagnate and pool up at any given individual.
Short term corruption doesn't pay, because accumulating excess money that you don't need via corruption just means your demurrage fee is higher.
Investing or lending money means giving it to someone that needs it more than you do and since the demurrage defeats the zero lower bound, it is rational to do so even at 0% interest since it allows you to avoid the demurrage.
And here is something that isn't intuitive for people whose mental model assumes permanent capital scarcity. If supply and demand on the money capital markets are balanced out, the expected interest rate is 0%.
Lower capital costs mean that products can be cheaper now. It is also much easier to invest into long term projects, e.g. against climate change, since the investment only needs to break even rather than growing the economy even further.
Oh yeah, also you now get full employment, like classical economic theory predicts. No more automation scare and feeling like the economy is trying to get rid of you or justify removing working class humans or being hostile to them.
The absence of unemployment makes welfare unattractive, which in turn means that governments will simultaneously have more tax payments and less welfare expenses simultaneously. All welfare should be paid out as demurrage currency in all countries. It's completely logical.
Since capital markets are now fair, it is possible to reduce total debt by paying off loans, whereas with permanent money the holder of money can always decide "I'll spend it later" thereby delaying the possibility of paying off a debt using that money, which is implicitly an extension of the duration of the debt. Given an infinite holding duration, the expected duration of the corresponding debt is obviously infinite too (eternal growing debt woooo).
Oh and it could have saved pension systems if it had been adopted ahead of time. Pension systems collapse because shrinking populations yield less output in the future but permanent money tells you that you can build a time machine to take labor from today and just teleport it straight into the future. Who needs young people working for you, if you can let the money work for you instead! Everyone knows the value of a dollar bill is backed by a humanoid robot inside the dollar bill that can perform exactly one dollar of work.
> ...instead of injecting $2 trillion and counting into a few AI companies...
Similar to the "why we spend money exploring space when there are hungry people on Earth?" question, I don't think this is a This Or That argument.
People and companies have different interests, some don't/can't care about education etc so they don't invest in those fields. Forcing these people/companies to invest in areas they don't interested in usually results in bad outcomes that is way worse than just let them be.
But some, do invest in education or civil infrastructure projects. It's not as hot, because... well, usually it makes less money from those things.
The core problem is still that, it is hard to figure out how to invest in such way that it could help the disadvantaged people, while at the same time maintaining a 10 or even 100x growth in the next 5 years.
From the company's perspective, there's no dilemma here, it's 100x growth potential ahead of everything else.
“This is how the world should work” I say once I read that the handful of entities engineering global economic calamity are privately held. I interrupt my chat bot girlfriend’s detailed but deeply incorrect summary of yesterday’s news to type “Only a literal child would want to go to a school.” “That is so true! You’re really on to something brilliant there! To get a head start before we drill down on this further I’ve deleted the contacts group ‘School Friends’.”
I don’t know what she’s talking about. I’ve never had a contact group with that title. Out the window my car is doing donuts on an old baseball diamond.
China is showing the world how it's done. Check out how many nuclear reactors they have built in the last 10 years. Turns out having a government that plans long term and actually delivers what they said they would makes a huge difference.
China glazing is strange. I think they are a cool country but they would have killed to be in the forefront in the AI race. There’s something obviously right that America is doing.
An amusing comment. Thanks. On a website, where every mention of China must be annotated with a disclaimer about how brutal and authoritarian they are, a comment that just states an opinion without it sure looks like "glazing". Let me know if it's safe to talk truthfully about George Floyd yet, free-man.
For context, the US alone spends annually a bit more than $2T on Social Security, Medicare, and Medicaid. That money is widely dispersed both geographically and socioeconomically. $2T is not as much as one may think. I think we should let investors put their private money where they wish.
Also, as an aside, the benefits of globalization on the balance far outweigh the drawbacks. Globalization has been the primary force pulling 3rd world countries out of poverty over the last 80 years.
I agree with you. The parent thinks 2 trillion is injected into AI companies. I think that is false and they are conflating it with market cap. It does make your point even better.
I don't think we don't really know how to do many of these things you list even if we have infinite funds. There is a real chance that we will mess things even more if we have infinite funds...
Of course we do, we were doing them until about the mid-70s and then the ultra-wealthy figured out how to game the system and we got “Greed is Good” Geckos running things since.
And, in case anyone needed proof, this is reflected in the US degree-completion-data, when measured as a percentage (https://galacticbeyond.com/two-percent-programmer/plots/over...), and when measured as a derivative of percentage (https://galacticbeyond.com/two-percent-programmer/plots/deri...). That green top-line, is business-majors, and those two lines that declined from top to average are social-sciences and education (all data from 1970 to 2011). In 1970 1 in 10 graduates were in business, 1 in 5 were in education and in social sciences. By 2011, 1 in 5 (or 2 in 10) were in business, and 1 in 10 were in social sciences and 1 in 20 were in education. Healthcare went from 3 in 100 to 1 in 10.
> and we got “Greed is Good” Geckos running things since
This phrase is the opposite of an exaggeration. It sounds like it should not be true, but it really, really is. To be fair though, if you told me in 2015 what the headlines for the 2020s would look like, I would assume you are some kind of satirist or comedian.
Always loved this from The Big Short (movie):
"In the years that followed, hundreds of bankers and rating agency's executives went to jail. The SEC was completely overhauled, and Congress had no choice but to break up the big banks and regulate the mortgage and derivatives industries.
Just kidding.
Banks took the money the American people gave them, and they used it to pay themselves huge bonuses, and lobby the Congress to kill big reform. And then they blamed immigrants and poor people, and this time even teachers. And when all was said and done, only one single banker went to jail."
Edit: I see I'm not the first to quote The Big Short in reply - such a good movie (and book)!
The narrative of the Big Short omits something big i.m.o.: encouragement of easy mortgage lending standards by Washington politicians. Is this line about the SEC the only reference to government in the movie?
Not this time. This time, we'll bail out them out because they are too big to fail and they need the money and assets. Next time we'll definitely get them. Pinky promise.
If it’s after the midterms, I’m doubtful. The AI leaders—apart from Dario—have gone particularly partisan. We also have a lot more post-crisis tooling that lets us wipe equity even when bailing out. See, for example, the ‘23 bank failures.
Even if one thinks we need to bail out the _companies_ to prevent more severe economic collapse, we could still arrest plenty of those who were in charge of these organizations.
It’s a different crowd this time. Instead of bankers and traders with fin instruments its Silicon Valley growth first sane economics later playbook just scaled up to dangerous levels
>"I have a feeling in a few years people are going to be doing what they always do when the economy tanks. They will be blaming immigrants and poor people." -Mark Baum, The Big Short
The scapegoats for this plan never change and have never changed in human history.
Increasingly more prominent is this new genre of activist hacker who thinks all big tech guys are bad parasites and if only they didn’t exist and we would have world peace and prosperity. That’s about the biggest virtue they know to signal.
The took the antithesis of the “great man” theory and ended up believing in a nonsensical version of it - every great thing in the world was done by normal people working 9-5. This is a kind of crude Labour Theory Of Value.
So you see these kind of comments signalling hatred in an attempt to show solidarity. But ultimately it is simply a signalling thing, a way to LARP as an activist because what else might one do as a simple rank and file employee. One wants to have a greater purpose to life and being an activist is a path for it.
Well assuming it's successful, there will be a large number of companies who's value will be reassessed as the token cost to replicate and run the business.
Multi-million dollar companies will be reduced into multi thousand dollar companies.
The CEOs will be replaced with teenagers in garages with their parent's credit cards.
If it stalls, then China will undercut the whole AI market with cheap electricity and crash the US stock market.
I can’t help but notice the “if it stalls” still assumes AI is successful, only that China beats the US. What if AI in general can’t do any of the things you mention?
It's obvious to more people every day that "AI" has never delivered on what it promised. Until now nobody cared because the costs were cheap and it pays to appear cutting edge. Now the price is rising, and the delayed cost of AI-addiction and broken outputs are starting to sting. Not that employers care about the individual harms of course, but a workforce so deep in collective delusion that they can't see the train coming is only useful when the market is delusional too.
Once this snaps, and it will snap suddenly, companies will be climbing over each other to rip out AI as fast as possible. They won't call it that, of course, you're not going to get a CEO on the news talking about how they made a mistake and it was wrong to invest so much in AI tech. But they'll mean it.
The win scenario is that the crash reduces "AI" use to near zero. Spat out into the graveyard of VC hype like blockchain and metaverse before it. Banished to an eternal unlife of scammers running call centre scams, deepfake porn producers, and the occasional "we made AI safe!" startup trying to reignite the bubble again. While companies with their business on the line clamber to announce that they don't use it.
no matter what, success or not the bubble is going to bust, it has inflated so much, the risk of it deflating slowly is a pipe dream. if AI doesn't turn into AGI, global financial crash. if AI turns into AGI and tons of people are out of work, global financial crash too.
> it has inflated so much, the risk of it deflating slowly is a pipe dream. if AI doesn't turn into AGI, global financial crash
Whether overvaluation can deflate gradually or suddently has to do entirely with debt and almost nothing to do with magnitude. AI is, currently, mostly equity financed.
I sold in May. I'm in the largest cash position (t-bills) I've ever had. I might be completely wrong. I just looked at the situation with the oil crisis, and without any obvious way to end it, I just figure, if there was every a more obvious "sell in May and go away" year, this is it.
Maybe I miss out on some gains and buy back in December or January. Maybe I'll increase my Berkshire position. I don't know. But companies tend to push the bad news till the end of the year, and boy, howdy there could be some bad news this year.
If we assume the play is the same as MAGA 1, the economic fallout happens post-2028. They'll keep inflating the bubble until MAGA cedes power. If that occurs.
Wishful thinking has it that we rally our representatives to let OpenAI and consorts rot. The last thing people should do is bail these delusional people out. Let them have it worse then WeWorks and let’s see if their self crowned AGI can help them out of their misery
We are 81 years away from the end of World War II. The baby boomers born after are the ripest target of financial sharks aiming to get a chunk of their retirements. People will be liquidating to settle estates, to pay inheritance taxes in large numbers. The AI boom feels like a stealthy rug-pull from other assets that are likely to tank from retirement withdrawals and into something that may last a little past the boomer assets wave.
Yes now that everyone’s equity is tied to overvalued assets, it’s a problem because it can have economy wide effects like in the subprime mortgage crisis.
"The five largest hyperscalers are set to spend over a trillion US dollars on AI-related capital expenditure from 2025 through 2026. These commitments are outpacing earnings and the free cash flow of these firms, leading some to issue debt to raise additional financing (Graph 11.A). This investment race may be partly driven by the perception that only a small number of players with superior technology will ultimately dominate the market shares. The
intense competition raises the risk of firms over-committing resources to investment projects with still uncertain returns, leaving all firms vulnerable to disappointments in AI payoffs. Model analysis based on such contest motives highlights the downside risk of current AI exuberance. As competitive pressure drives capex higher, the net economic surplus – the total payoff less investment costs – declines for the sector as a whole and could turn negative in adverse scenarios (Graph 11.B). Disappointment in returns could trigger a sudden pullback in financing and turn the capex boom into a protracted investment bust, with potential knock-on effects on financial conditions (see below).
Another risk is that the AI boom runs into a supply side roadblock. The AI build- out has recently been facing growing bottlenecks in electricity, advanced semiconductors and grid equipment. Fast-growing demand for computing power is already pressuring electricity prices and input costs, with potential spillovers to inflation. Looking ahead, these temporary shortages may also amplify over-investment, as firms attempt to lock in future capacity through long-dated contracts that further expose them to any disappointments in demand.
...
Should inflation rise significantly or AI-led investment turn to a bust, the macroeconomic consequences could be amplified by existing financial vulnerabilities. A tightening of policy rates needed to contain inflation could precipitate a sharp pullback in asset prices after a prolonged period of exuberant risk-taking, triggering
disruptive macro-financial feedback loops. A reversal of AI optimism could likewise have major financial consequences, given AI firms’ rising leverage and growing footprint in credit markets. Vulnerabilities extend to their supplier ecosystem, including engineering, procurement and construction (EPC) contractors whose balance sheets are comparatively weak, leaving them exposed to any capex pullback by hyperscalers.
...
A sharp repricing of equity risk could prompt a reassessment of corporate credit risk and lead to tighter credit conditions more broadly.1 Indeed, broad indices of credit spreads tend to correlate negatively with stock market returns (Graph 14.A), more so for the high-yield than the investment grade segment. While large, synchronised corrections in both markets are rare, there are notable precedents such as the Great Financial Crisis and the March 2020 dash for cash episode. A repricing of risk this time, whether triggered by higher interest rates or an AI bust, has the potential to be similarly disruptive by triggering a corporate credit freeze with wider implications for aggregate investment."
Pretty much. Given the levels of investment in capital and research if AI companies actually hit what they're aiming at they'd have to collapse the labor market to recoup, bricking the economy in the process. Given the levels of outside investment inflating valuations, if the bubble pops it's 2008 all over again. There's this incredibly narrow window of "just useful enough to extract rents" where everything doesn't go to shit.
No, it's correct. The best (short-term) case is that they become eternal parasites. If they fail to do that, they'll bring a lot down with them when they fall.
>The best (short-term) case is that they become eternal parasites.
Producing a product that delivers value and people are willing to pay for makes you a "parasite"? Sure, it might cause massive disruptions to the labor market, but that's mostly orthogonal to whether it's a "parasite" or not. Mechanized farming has almost wiped out agricultural employment (compared to pre-industrial levels), but that doesn't make tractor manufacturers or fertilizer companies "parasites"
tractor manufacturers or fertilizer companies didn't suck down the work of generations of predecessors in a questionably-legal fashion only to turn around and sell a heavily discounted version of that back to them. I'm not sure where "parasite" becomes appropriate, but your analogy is poor.
Maybe in the eyes of seething artists/programmers seeing their jobs getting automated, but courts have so far ruled that AI training falls under fair use.
Moreover it's not hard to think of vaguely similar objections to fertilizers. They're often produced at some harm to society, as well as their use. They're also in some sense, a "heavily discounted" versions of that they replaced, bird guano or whatever.
> Moreover it's not hard to think of vaguely similar objections to fertilizers.
It's completely different. If LLM companies pulled this out of thin air it would be also different, but no; they've effectively plundered the commons and locked up all the profit for themselves. If intellectual labor goes the way of agricultural labor, I think humanity will have lost something valuable.
And don't come back with the "farmers would have said the same thing about the industrial revolution!" thing again if you're just going to terminate your thought there. Automating agricultural labor brings vast material benefits for all since it lowers the cost of tangible goods needed for life. I'd challenge you take this one step further and explain why automating intellectual labor will provide similar fruits and is therefore something to cheer for.
If we actually do meaningfully automate intellectual labor, we create a world where we have real technical solutions for our toughest problems. Maybe we can get carbon capture and fusion energy working. There’s a theoretical world of abundance for us to explore.
That’s the steel man argument.
FWIW I mostly don’t believe that LLMs are the answer, I don’t think they’re going to reach a high enough level of capability to do this, and I think the current AI companies are problematic in a lot of ways.
I also think LLM use is bad for us and probably harms our thinking abilities. And using it takes away a lot of what it means to be human.
Personally I like both physical and mental difficulty. I like gardening even if I could just buy mass produced flowers. I like riding a bike even though cars are “easier”. I like playing ukulele with my family even though I can barely make a chord, much better than listening to some other real musician, or Suno ai generated songs. I like eating my wife’s sourdough bagels even if they take several hours more than just buying some.
And I think having those regular challenges and achievements make life worth living! And I worry that the AI future that some envision will make much of what we get value from feel meaningless in the same way that writing code by hand is starting to.
Maybe we’ll still be fine in the same way I find meaning in all of those things that I listed above. But damn what a gamble
No, that's actually more destructive in the short term. That was the OP's point. If they actually become something effective, they will utterly destroy the economy. If they fail, they will drag the economy down around them as they go. The least destructive possible outcome is becoming eternal parasites.
Well there's also the fact that fundamentally and ultimately, AI is incompatible with the economic system. Capitalism is rooted in human labour having positive economic value, and hence demand. AI will ultimately automate all labour, making the economic value 0. Eventually capital generation will simply die and the system crashes.
Yeah but the investments arent aiming for churning out SaaS apps. Its to automate large swathes of intellectual labour. Of which only SWE has been cracked yet. There is a question mark as to if the others will crack. If they arent then these investments will collapse from speculation down to reality. That possibility is what is being discussed here
As to whether that will happen, I think that risk is real. Because claude code isnt made by the generalozed capabilities of the tech but by good old non-generalozable hueristics and rule based engines. I dont think that will scale to other feilds at the factor these investments assume. Its the bitter lesson again. It scales with deliberate and specific design, not data, so it wont scale
We learnt this with ibm watson. Deepblue achieved chess supremacy but the last mile wasnt data driven, it was heiristic driven, and so watson, its successor, couldnt scale/generalize.
My prediction is that this speculation on LLMs with harnesses will collapse since they wont scale. We'll have another winter where the reasearchers will be leaft alone long wnough to come up with the next breakthrough (probably game theory based data driven agency) which might then create what this hypecycle is speculating
There’s argument to be made that SWE hasn’t been cracked either. The latest models are great at coding medium sized applications, but figuring out the requirements and consolidation of domain knowledge is something still lacking
you mentioned a very good point about scalability. we're seeing alot of productivity gains, but only from SWEs, which are but a very small segment of the global economy. all other economic use cases require thorough last-mile development and iteration that is not too different with current automation tools.
A friend who is a psychologist was telling me he thinks in another year or two insurance companies will insist people see an AI therapist first before being willing to pay for a real person.
LOL, the same AI that has landed companies in court defending themselves against wrongful death lawsuits for helping someone convince themselves suicide is the right answer, and even encouraging them? That AI? I am unclear that any insurance company is going to want a piece of that action anytime soon.
What you've just told me is that psychologists, just like SWEs, are prone to thinking they know how business works but in fact know fuck all.
Tracks. This is that same speculation that AI will be good enough. Wonder what a crash in that isecase will look like? Increased suicide rates? More instances of psychosis? It might not wven be directly measurable or easily traceable to AI therapists. Would suck
All those automaton tools will eventually be initially one-shotted and then monitored by LLMs though. There probably won't be a "last mile" per se; just constant tweaking and optimizations throughout, within a feedback loop.
What your describing is iterating in the last mile. Your assuming that the AI will iterate in the last mile with the same efficacy that it iterates before that. I think that will fail. Thats the bitter lesson, that adhoc solutions to the last mile (rather than generalozed solutions that scale with training data) asymtotally stall and so dont scale.
Meaning claude code wont be able to make a "claude video editing" or "claude accounting" with the current tech. Human experts will need to encode their knowledge into it for the last mile and that wont scale the way these speculations expect
We aren't seeing productivity gains in software either. What we are seeing is a lot of people who claim to be more productive, but in fact are building piles of tech debt that will fall over before long. But hey, they're building that tech debt really fast!
> but in fact are building piles of tech debt that will fall over before long
This is speculation as well. Its well founded but speculation nonetheless. Youre speculating things will stay the way they have till now.
I do see your point but what makes me consoder the other side is that ive been building an app that reaches ~10k LOC, purely with opus, no code review at all, and it hasnt hit any tech debt issues that i havent easily been able to address. Setting up good context management meant that claude could just figure things out itself.
And for reference this is an app that manages an ethernet camera, runs vehicle detection on the stream, and surfaces the detections on an ipad for operators to inspect and annotate for cellphone usage, so not trivial. Needed good architechtong and design from my end, but it was honestly scarily easy. So idk what the threshold for tech debt crash is but it wasnt there
We will find out how much of work is given to people just so that there's a person/company associated with a technical decision. I personally think this might be quite high.
Exactly. I build automation tools for my company which have improved productivity quite a lot and put precisely zero people out of a job. Partly we find other things for them to spend time on, and partly it just turns out that we like to have humans doing jobs.
It is cliche at this point that HN is the place you go to hear software developers reduce all of the world's problems into simple algorithmic arguments which for some reason never actually solve anything. Not shocked that we are similarly incapable of understanding that algorithmically replacing a software developer isn't easy just because we think we know what the job is.
While I don't agree with your premise at all, even if it could one shot a SaaS product (a statement so vague it's meaningless) I don't think there's much of an argument for why that's economically useful. A lot of SaaS has free software/open source equivalents anyway (how else do you think the clanker's are able to plagiarize it?). People still pay for Office even though you could easily use LibreOffice, or GitHub when you could self host Forgejo. It's like when anthropic made a big deal out of making a broken compiler. Neat, so, after ingesting all of open source and burning a trillion tokens you ended up with something worse than what's already out there; and instead of doing something economically useful like giving a person money to build it or supporting the open source ecosystem, you're just wasting energy on datacenters.
So good at it that I’m right now in the process of building instead of buying.
Here’s how that plays out in the economy:
- My company spent $50 on my tokens to build this internal tool
- Anthropic spent $XXX to deliver those tokens to me.
- The company I was going to buy the tool from lost $XX,XXX per year that I would have paid them.
I dunno, kind of sounds like the economy just got smaller.
I could usually accept the idea that software getting cheaper generally increases demand for software and expands the economy surrounding it, but I’m not sure if we have precedent for what happens when software becomes positively worthless.
I certainly can’t give you a better answer for my company than “it depends” or “I don’t really know.”
The company could just be happy to have better margins and be happy the stock finally went up. It might literally do nothing with them or do something economically unproductive like buy back stock.
What I can tell you with certainty is that we aren’t going to hire anyone else or launch any other product as a result. Our business just isn’t at that level of growth potential.
Perhaps we can surmise that money going to shareholders can grow the economy. They’ve got more money to reinvest in other stuff.
But then again, if everyone can shart out a SaaS app with $50 in tokens, what software companies will they want to invest in?
AI gives me that feeling of “what happens to bakers and butchers when the supermarket gets invented and they decide to sell bread and meat at or below cost?”
Every company has a list of >WACC IRR projects that it can spend saved money on. If not, it’s a cash cow company that wasn’t growing in the first place and will allow shareholders to use the saved cash for other economically expanding projects.
What companies can expand if the income of consumers is shrinking. This is the scary bit to me — AI crashes and takes the economy with it, or; AI succeeds as promised and people go unemployed and crash the economy.
The only path that isn’t disastrous is threading the needle of “just right” productivity gains. The people in charge aren’t smart enough to give me warm fuzzy feelings on that.
Economy != software companies. Maybe there needs to be a capital rotation out of tech? That’s speaking beyond my expertise though (and imo is a little too doomer). Continuing the hypothetical through: Healthcare, industrials, financial services and many other verticals have plenty of growth opportunities.
Otherwise it would probably be the software companies that are the most focused on last-mile details (where AI in my experience has the most trouble). I expect that as consumers are faced with more and more AI slop SaaS they will be increasingly willing and able to pay for quality.
And if the company didn't need $XX,XXX*0.90 (or more) that you would have paid them to further develop their product and stay in business? If that other company now paid their own $50 in tokens? Maybe the overall flow of money in the economy went from $XX,XXX (you) + $XX,XXX (them) + $(not much, AI didn't exist yet) equals or is greater to $50 + $50 + ($xbillions in AI)? Dunno.
wouldn't take long for them to find out with the developer bragging about it. he one-shot vibe-coded a sass app (keyword: one-shot), that says alot about the quality.
> Meanwhile AI has gotten so good it can just about one shot a SaaS app.
There isn't a direct correlation between AI improvement or stagnation and whether or not the amount being spent by AI labs and the associated ecosystem will result in a financial crash.
Look into the history of railroads and the internet itself to see how massive levels of investment can result in economic crashes even when the thing being invested in produces real, widespread societal value.
One could argue that one of the nightmare economic scenarios for AI is actually that it gets too good too fast and results in a wipeout of the white collar worker that we are currently nowhere near ready to deal with given how propped up our economy is on consumer spending.
difference this time is they have "fiat money" and money printer. Market and all inv. bankers knows that in major crash they will print unlimited amounts so back to same prices or near them. printer is still printing and it's only goes to selected investments
As if they did not back then. Fiat is just simpler to work with, but one can pull a bubble without it just fine. Anyone forgot the railroad crash of 1873? The tulip mania of 17th century?
I'm not usually for arguments of "this money could have been better spent elsewhere", but here's a thought experiment. Lets say instead of injecting $2 trillion and counting into a few AI companies, we instead injected $2 trillion dollars into things like infrastructure (real infrastructure, not GPU warehouses), education, helping out communities ravaged by globalization (I doubt most people on Hacker News venture outside of coastal areas, but if you want to make a REAL difference as an entrepreneur why not look at parts of the country that are struggling and figure out how you could make a difference there? You know, instead of trying to just ruin the economy for the sake of the already-obscenely-wealthy). I'm not saying all those ventures would succeed, but I think that amount of money put towards boring-but-real problems would make a much bigger positive impact for everyone.
This is 100 manhattan projects. Lots and lots and lots of things that can be done. Reminds me of the quote "water water everywhere not a drop to drink" ==> Money, money everywhere, not a cent goes towards something remotely useful.
1) Could've added nifty features to roads (rf beacons?) and make it super easy to build autonomous vehicles. This would've solved so many problems, but most importantly 45,000 lives every year and millions of accidents.
2) Could've given the money to farmers who are getting ethanol subsidies today to add panels, supply tons of power to the grid, creating permanent income to the farmers instead of handouts every year.
3) Could've built many Universities permanently free which is feasible with an endowment that large.
4) Could've built walking/biking infrastructure in most cities, which would help reduce healthcare expenses. Most of the expenses today are lifestyle/metabolic diseases.
5) Could've switched the entire country to clean energy, exporting ALL of the oil and create a sovereign wealth fund.
6) Could've built a city for homeless people with homes, excess energy production (solar) providing the function of a UBI, a massive hospital with health and mental health facilities, de-addiction centers. New city in the middle of nowhere where land is dirt cheap. Army corps of engineers could execute this.
7) Rehabilitate prison (not like George Carlins suggestion of Kansas: https://www.youtube.com/watch?v=PkMrFIv_QHk), but more like nordic countries.
8) Build solar over all the interstates.
You are describing altruistic gifts, not investment meant to generate a return.
All this money is being deployed to try and turn a dime into a dollar.
The biggest area I can see where money could be deployed into building something useful -- while generating a positive return -- is real estate development in cities with massive housing shortages. That's basically not allowed in the same cities. In large part, because people who yearn for an more altruistic world care very much about their own quality of life and property values than they do about the well-being of their community in the long run.
It doesn't generate private returns but investing in your worker population definitely nets returns for the public.
Also, some might say that upending the de facto copyright regime in favour of AI companies was an altruistic gift.
Typically the best return on investment is to build proper separated active travel infrastructure (e.g. cycle lanes).
From https://www.cyclinguk.org/briefing/case-cycling-economy:
> For every £1 invested, walking and cycling return an average of around £5-6. A ‘benefit to cost ratio’ (BCR) greater than or equal to 4 is considered to be ‘very high’ value for money. Putting this in some context, the predicted BCR for the HS2 (high speed railway) wasn’t nearly so impressive at about £2.30 for every £1 invested.
It's private money. At least in my country the government isn't giving a single cent to AI outside of fundamental research projects. If companies want to bet all their money at roulette they are free to do so.
Its private money now. Then they get bailed out.
My understanding of injecting money in education is that it's proven to be extremely ineffective at improving outcomes.
Schools just hire more administrators and build nicer gyms.
You have to wait 20 years for the returns to society. Public education was enormously successful when it was introduced in the 19th century. There's just no profit in waiting for second order effects to kick in.
What makes you choose the 19th century for public education creation? It was my understanding the Puritans and thusly Yankees that spread through NE and the north Midwest were staunch proponents of schoolhouses and publicly funded education dating back to the to the early 17th century.
So was Rural Free Delivery. Farmers being able to communicate was a massive boon. There is a channel for farmers called RFD tv. They completely scrubbed the free provided by the government part after private equity bought the tv channel targeting farmers. Then they got Imus in the Morning so farmers listed to Imus, Rush, Hannity, and orielly forgetting the government helps them.
Diminishing returns. Per-student education spending has been going up since 1990 except a dip during the 2008 recession. Adjusting for inflations it’s now double what it was 30 years ago.
The population of students is shrinking and there is (unnecessarily) growing overhead that has to be paid for.
The people in charge of the schools don’t seem to think it’s unnecessary overhead?
of course their luxury SUVs wouldn't think it's an unnecessary overhead. Come on guy, those expense accounts aren't going to pay for themselves...
chop chop... get to it.
Why would they be any better than the people distributing the money to them in the first place? The students working hard at "providing the good educational outcomes", by this economy's standards, are suckers.
Investment in education is one of the best ROI a society can have. Quite surprising that in the USA this relationship has been eroded by schools spending in sports and admin, in most other developed countries it's a quite direct relationship between more investments -> better educational outcomes.
But there's 2 problems with that. To save money governments declare that "teaching is a profession". By which they mean: "teaching", not subject matter expertise, matters. Except:
1) the single thing you cannot beat is capable teachers. And NOT capable as in able to teach, subject-matter-experts. And that is what everybody wants to avoid at all costs.
Put it like this: an unwashed, abrasive, offensive smelly asshole with a master in math can teach kids math. A super-friendly, nice, beautiful, polite social science major (or not even that) who doesn't know math well cannot. The level of teachers needs to be an order of magnitude above the level expected of kids at the end of the class. Knowledge of the subject needs to take absolute precedence in teacher hiring decisions, to the point that good teachers get rejected if lacking expertise in the subject.
This is not how things work. They only worked that way for a short while due to the crash of 89. I've had a French teacher that couldn't speak more than "some" French. Certainly nowhere near fluent.
2) Money needs to support that. Which means for Math, the money for math teachers needs to be competitive with research departments of banks, and very different for different subjects, which seems offensive to schools and government in general. If teaching positions aren't competing with other professional careers in the subject matter, it will never work. This will, of course, mean that many science teachers make more than a school principal and even a district administrator. Deal with it.
For Math, they are literally not even 25% of that, but even for French it's insufficient.
> 2) Money needs to support that. Which means for Math, the money for math teachers needs to be competitive with research departments of banks, and very different for different subjects, which seems offensive to schools and government in general. If teaching positions aren't competing with other professional careers in the subject matter, it will never work. This will, of course, mean that many science teachers make more than a school principal and even a district administrator. Deal with it.
I believe this to be a heavily US-centric view, Estonia and Finland do not pay more for math teachers than they could get in banks' research departments but still manage to get competent teaching across most school subjects.
Things work outside of the US, the issue is the US does not want to look outside and adapt itself to incorporate what's been learnt from other systems. It's a "do the US way or no way" attitude, and it's slowly eroding to become more a "no way" than any kind of "US way" of doing things.
Doesn't change the fact that I have 2 daughters that are getting Math education from someone who is easily beat in math teaching quality by Google Gemini (I realize this is a specific strong point of Gemini, but still)
And I do mean easily beat.
Don't forget religion! Some crazy things being taught at American schools that you couldn't get away with in secular nations.
Well yes, you have to spend the money wisely. How could we construct a system so that we have 2x as many teachers (thereby halving the classroom size)? That would have a lot of good second-order effects beyond test scores.
So why has per-capita student spending doubled since 1990 (adjusted for inflation) without any increase in test scores? Why haven’t we been spending the money wisely?
Student to teacher ratios have continuously decreased and are about half of what they were in 1960. Data on the results is mixed: https://www.brookings.edu/articles/class-size-what-research-...
The funding system is broken, not the actual act of funding.
We aren't incentivizing meaningful work, we are incentivizing diplomas.
Because we have also increased the spending in "un-education" (entertainment, social media, college sport...) ?
What's your own theory ?
Honestly curious, but don’t tests also adjust for “inflation”? Aren’t tests today be harder than they were in 1990?
See https://www.oecd.org/en/data/indicators/mathematics-performa...
Maybe don't just 'inject' it.
Maybe use it to increase outcomes.
Trying to micromanage how the money gets spent is how you get administrators.
What about into research grants?
> we instead injected $2 trillion dollars into things like infrastructure (real infrastructure, not GPU warehouses), education, helping out communities ravaged by globalization
Even excluding military spending, US governments spend $2 trillion every 10 weeks.
Just to be clear: you're talking about federal and state non-military spending.
And about 10% of this is interest. So over the course of a year, the US is paying about $1.25 trillion in interest at the federal and state level.
Why wouldn’t you include state spending? That’s the level of government primarily responsible for infrastructure and education.
I wasn't making a judgment about including or not including state spending. It's just that "US governments" is not a common way for Americans to describe federal and state. People think federal when they see "US government".
Gotcha. Was being lazy and typing on my phone, sorry.
How? The annual federal budget is roughly $7T.
You have to include state and local spending too. We’re at 40% of GDP which works out to almost $13 trillion: https://fred.stlouisfed.org/graph/?g=1CFpQ. Subtract $1 trillion in defense, and we’re spending about $1 trillion a month on government.
Does that chart double-count state transfers to municipalities? When I was in local government, about half our school budget came from the state, so there would be entries on both ledgers.
> governments
Plural implies they count more than the federal government.
Not to mention that data centres are infrastructure!
Other nations are falling behind and will be at a real disadvantage soon.
AI datacenters don't really confer any geographical benefit since their output can be transmitted at extremely low cost. The benefit of local datacenters is purely political and has nothing to do with technological advantage.
People invest in what would get them the highest ROI. No one is really thinking about “improving the world”* when they invest.
What’s scary about the AI boom is people over investing and not being able to recoup their investment which will lead to knock on effects - companies going bankrupt and people losing jobs, savings gone, … etc.
* As ESG has shown, not everyone agrees on what is considered “improving”.
It won't make much difference. The US has a lot of problem, but "not spending enough money" isn't one.
The US government spend a lot on healthcare ($5.3 in 2024)[0]. More than most European countries per capita. But many people still feel that the US hardly has healthcare at all. Pouring more money without a full structural overhaul will likely make things worse.
And the $2T you mentioned is investors' money, which means that your plan is actually to increase tax by $2T and pour it into a system proven inefficient.
[0]: https://www.healthaffairs.org/doi/10.1377/hlthaff.2025.01683
It all sounds great, but unfortunately human nature is that money attracts corruption, so how could a $2T injection be managed in a way that ensures everything is square, without adding significant overhead to the spending?
Governments are often just as bad at this as private entities are.
If white collar crime was actually punished then this wouldn't be such a regular occurrence
Mostly this, everywhere, going back to every time.
It's quite easy. The money you "inject" needs to have demurrage.
You can then give all of the $2 trillion demurrage dollars directly to the most corrupt individual on the planet and it won't matter.
There are two scenarios:
1. The recipient keeps the money: It will automatically be returned. 2. The recipient spends the money: It has flown from the most corrupt individual on the planet to a less corrupt individual.
If we assume that each transaction allows the next recipient to make a decision on how to best spend the money, it would automatically spread out, since it cannot stagnate and pool up at any given individual.
Short term corruption doesn't pay, because accumulating excess money that you don't need via corruption just means your demurrage fee is higher.
Investing or lending money means giving it to someone that needs it more than you do and since the demurrage defeats the zero lower bound, it is rational to do so even at 0% interest since it allows you to avoid the demurrage.
And here is something that isn't intuitive for people whose mental model assumes permanent capital scarcity. If supply and demand on the money capital markets are balanced out, the expected interest rate is 0%.
Lower capital costs mean that products can be cheaper now. It is also much easier to invest into long term projects, e.g. against climate change, since the investment only needs to break even rather than growing the economy even further.
Oh yeah, also you now get full employment, like classical economic theory predicts. No more automation scare and feeling like the economy is trying to get rid of you or justify removing working class humans or being hostile to them.
The absence of unemployment makes welfare unattractive, which in turn means that governments will simultaneously have more tax payments and less welfare expenses simultaneously. All welfare should be paid out as demurrage currency in all countries. It's completely logical.
Since capital markets are now fair, it is possible to reduce total debt by paying off loans, whereas with permanent money the holder of money can always decide "I'll spend it later" thereby delaying the possibility of paying off a debt using that money, which is implicitly an extension of the duration of the debt. Given an infinite holding duration, the expected duration of the corresponding debt is obviously infinite too (eternal growing debt woooo).
Oh and it could have saved pension systems if it had been adopted ahead of time. Pension systems collapse because shrinking populations yield less output in the future but permanent money tells you that you can build a time machine to take labor from today and just teleport it straight into the future. Who needs young people working for you, if you can let the money work for you instead! Everyone knows the value of a dollar bill is backed by a humanoid robot inside the dollar bill that can perform exactly one dollar of work.
> but if you want to make a REAL difference as an entrepreneur why not look at parts of the country that are struggling
Why not parts of the world even? There are more countries than "the country". Some of us here aren't even from "the country".
> ...instead of injecting $2 trillion and counting into a few AI companies...
Similar to the "why we spend money exploring space when there are hungry people on Earth?" question, I don't think this is a This Or That argument.
People and companies have different interests, some don't/can't care about education etc so they don't invest in those fields. Forcing these people/companies to invest in areas they don't interested in usually results in bad outcomes that is way worse than just let them be.
But some, do invest in education or civil infrastructure projects. It's not as hot, because... well, usually it makes less money from those things.
The core problem is still that, it is hard to figure out how to invest in such way that it could help the disadvantaged people, while at the same time maintaining a 10 or even 100x growth in the next 5 years.
From the company's perspective, there's no dilemma here, it's 100x growth potential ahead of everything else.
because that's not how this world works, and neither it should. most people figure that out in their early teens.
and fyi, that 2t is not tax money, it's someone's money.
“This is how the world should work” I say once I read that the handful of entities engineering global economic calamity are privately held. I interrupt my chat bot girlfriend’s detailed but deeply incorrect summary of yesterday’s news to type “Only a literal child would want to go to a school.” “That is so true! You’re really on to something brilliant there! To get a head start before we drill down on this further I’ve deleted the contacts group ‘School Friends’.”
I don’t know what she’s talking about. I’ve never had a contact group with that title. Out the window my car is doing donuts on an old baseball diamond.
>my chat bot girlfriend
you may think she's just your gal but she may be everyone's pal.
llm's are the town bicycle
And why’s that?
China is showing the world how it's done. Check out how many nuclear reactors they have built in the last 10 years. Turns out having a government that plans long term and actually delivers what they said they would makes a huge difference.
China glazing is strange. I think they are a cool country but they would have killed to be in the forefront in the AI race. There’s something obviously right that America is doing.
An amusing comment. Thanks. On a website, where every mention of China must be annotated with a disclaimer about how brutal and authoritarian they are, a comment that just states an opinion without it sure looks like "glazing". Let me know if it's safe to talk truthfully about George Floyd yet, free-man.
For context, the US alone spends annually a bit more than $2T on Social Security, Medicare, and Medicaid. That money is widely dispersed both geographically and socioeconomically. $2T is not as much as one may think. I think we should let investors put their private money where they wish.
Also, as an aside, the benefits of globalization on the balance far outweigh the drawbacks. Globalization has been the primary force pulling 3rd world countries out of poverty over the last 80 years.
You are comparing flow with a static amount. 2 trillion is the market cap of AI companies right?
No I am not. The discussion is about investment spending on AI.
It’s actual less than $2T IIRC, which makes my point even more.
I agree with you. The parent thinks 2 trillion is injected into AI companies. I think that is false and they are conflating it with market cap. It does make your point even better.
I don't think we don't really know how to do many of these things you list even if we have infinite funds. There is a real chance that we will mess things even more if we have infinite funds...
Of course we do, we were doing them until about the mid-70s and then the ultra-wealthy figured out how to game the system and we got “Greed is Good” Geckos running things since.
And, in case anyone needed proof, this is reflected in the US degree-completion-data, when measured as a percentage (https://galacticbeyond.com/two-percent-programmer/plots/over...), and when measured as a derivative of percentage (https://galacticbeyond.com/two-percent-programmer/plots/deri...). That green top-line, is business-majors, and those two lines that declined from top to average are social-sciences and education (all data from 1970 to 2011). In 1970 1 in 10 graduates were in business, 1 in 5 were in education and in social sciences. By 2011, 1 in 5 (or 2 in 10) were in business, and 1 in 10 were in social sciences and 1 in 20 were in education. Healthcare went from 3 in 100 to 1 in 10.
> and we got “Greed is Good” Geckos running things since
This phrase is the opposite of an exaggeration. It sounds like it should not be true, but it really, really is. To be fair though, if you told me in 2015 what the headlines for the 2020s would look like, I would assume you are some kind of satirist or comedian.
Just return the taxes instead
I mean, you want to be annoyed at something, be annoyed at Apple for investing 3x the Marshall plan into China, instead of America.
Surely this time we'll learn our lesson and disempower the parasites that create these situations, right? Right guys?
Always loved this from The Big Short (movie): "In the years that followed, hundreds of bankers and rating agency's executives went to jail. The SEC was completely overhauled, and Congress had no choice but to break up the big banks and regulate the mortgage and derivatives industries.
Just kidding.
Banks took the money the American people gave them, and they used it to pay themselves huge bonuses, and lobby the Congress to kill big reform. And then they blamed immigrants and poor people, and this time even teachers. And when all was said and done, only one single banker went to jail."
Edit: I see I'm not the first to quote The Big Short in reply - such a good movie (and book)!
The narrative of the Big Short omits something big i.m.o.: encouragement of easy mortgage lending standards by Washington politicians. Is this line about the SEC the only reference to government in the movie?
Not this time. This time, we'll bail out them out because they are too big to fail and they need the money and assets. Next time we'll definitely get them. Pinky promise.
> we'll bail out them out
If it’s after the midterms, I’m doubtful. The AI leaders—apart from Dario—have gone particularly partisan. We also have a lot more post-crisis tooling that lets us wipe equity even when bailing out. See, for example, the ‘23 bank failures.
People are already sour on the economy. They are going to be in a whole mood if we get a real, serious recession.
Even if one thinks we need to bail out the _companies_ to prevent more severe economic collapse, we could still arrest plenty of those who were in charge of these organizations.
It’s a different crowd this time. Instead of bankers and traders with fin instruments its Silicon Valley growth first sane economics later playbook just scaled up to dangerous levels
Next time it’ll be different though
>"I have a feeling in a few years people are going to be doing what they always do when the economy tanks. They will be blaming immigrants and poor people." -Mark Baum, The Big Short
The scapegoats for this plan never change and have never changed in human history.
When the world goes to shit society does two things: search for a strong leader and build temples.
Dying with dignity is never in the cards.
Which parasites?
The hoarders.
Elon Musk, Sam Altman, Jeff Bezos, Mark Zuckerberg, etc
Yes, and then you'll build a communist paradise. It's just one purge away.
Increasingly more prominent is this new genre of activist hacker who thinks all big tech guys are bad parasites and if only they didn’t exist and we would have world peace and prosperity. That’s about the biggest virtue they know to signal.
The took the antithesis of the “great man” theory and ended up believing in a nonsensical version of it - every great thing in the world was done by normal people working 9-5. This is a kind of crude Labour Theory Of Value.
So you see these kind of comments signalling hatred in an attempt to show solidarity. But ultimately it is simply a signalling thing, a way to LARP as an activist because what else might one do as a simple rank and file employee. One wants to have a greater purpose to life and being an activist is a path for it.
Well assuming it's successful, there will be a large number of companies who's value will be reassessed as the token cost to replicate and run the business.
Multi-million dollar companies will be reduced into multi thousand dollar companies.
The CEOs will be replaced with teenagers in garages with their parent's credit cards.
If it stalls, then China will undercut the whole AI market with cheap electricity and crash the US stock market.
So what exactly is the win scenario here?
I can’t help but notice the “if it stalls” still assumes AI is successful, only that China beats the US. What if AI in general can’t do any of the things you mention?
It's obvious to more people every day that "AI" has never delivered on what it promised. Until now nobody cared because the costs were cheap and it pays to appear cutting edge. Now the price is rising, and the delayed cost of AI-addiction and broken outputs are starting to sting. Not that employers care about the individual harms of course, but a workforce so deep in collective delusion that they can't see the train coming is only useful when the market is delusional too.
Once this snaps, and it will snap suddenly, companies will be climbing over each other to rip out AI as fast as possible. They won't call it that, of course, you're not going to get a CEO on the news talking about how they made a mistake and it was wrong to invest so much in AI tech. But they'll mean it.
The win scenario is that the crash reduces "AI" use to near zero. Spat out into the graveyard of VC hype like blockchain and metaverse before it. Banished to an eternal unlife of scammers running call centre scams, deepfake porn producers, and the occasional "we made AI safe!" startup trying to reignite the bubble again. While companies with their business on the line clamber to announce that they don't use it.
https://archive.is/xMD3t
why am i looping back in the captcha when confirming?
I think that’s what happens if you’re on Cloudflare DNS, due to a nerdy dispute they’re having with archive.is.
Is there any comprehensive list of historical warnings from central bankers?
I just checked 2007 and 1999 reports [1] a a bit and doesn't seem like they made such obvious warnings at those times.
I don't know much about economy and I just did some ctrl + f skimming, but this new 2026 warning is obviously more clear to me.
[1] https://www.bis.org/annualeconomicreports/index.htm?annualec...
You're on the spot, it can certainly be a self-defeating prophecy
https://en.wikipedia.org/wiki/Self-defeating_prophecy
I think we also would need to know how many of these warnings they gave where nothing bad happened
no matter what, success or not the bubble is going to bust, it has inflated so much, the risk of it deflating slowly is a pipe dream. if AI doesn't turn into AGI, global financial crash. if AI turns into AGI and tons of people are out of work, global financial crash too.
> it has inflated so much, the risk of it deflating slowly is a pipe dream. if AI doesn't turn into AGI, global financial crash
Whether overvaluation can deflate gradually or suddently has to do entirely with debt and almost nothing to do with magnitude. AI is, currently, mostly equity financed.
How do you prepare for such thing? Do you sell all stocks now? Then what?
I sold in May. I'm in the largest cash position (t-bills) I've ever had. I might be completely wrong. I just looked at the situation with the oil crisis, and without any obvious way to end it, I just figure, if there was every a more obvious "sell in May and go away" year, this is it.
Maybe I miss out on some gains and buy back in December or January. Maybe I'll increase my Berkshire position. I don't know. But companies tend to push the bad news till the end of the year, and boy, howdy there could be some bad news this year.
If we assume the play is the same as MAGA 1, the economic fallout happens post-2028. They'll keep inflating the bubble until MAGA cedes power. If that occurs.
Wishful thinking has it that we rally our representatives to let OpenAI and consorts rot. The last thing people should do is bail these delusional people out. Let them have it worse then WeWorks and let’s see if their self crowned AGI can help them out of their misery
My guillotine & rope startup is going to make a killing (no pun intended).
We are 81 years away from the end of World War II. The baby boomers born after are the ripest target of financial sharks aiming to get a chunk of their retirements. People will be liquidating to settle estates, to pay inheritance taxes in large numbers. The AI boom feels like a stealthy rug-pull from other assets that are likely to tank from retirement withdrawals and into something that may last a little past the boomer assets wave.
"Claude, how do I become a prepper?"
Step 1. Get a Costco or Sam's Club membership
Step 2. Buy COST stock on margin.
same as it ever was
Yes now that everyone’s equity is tied to overvalued assets, it’s a problem because it can have economy wide effects like in the subprime mortgage crisis.
The BIS report: https://www.bis.org/publ/arpdf/ar2026e1.pdf
"The five largest hyperscalers are set to spend over a trillion US dollars on AI-related capital expenditure from 2025 through 2026. These commitments are outpacing earnings and the free cash flow of these firms, leading some to issue debt to raise additional financing (Graph 11.A). This investment race may be partly driven by the perception that only a small number of players with superior technology will ultimately dominate the market shares. The intense competition raises the risk of firms over-committing resources to investment projects with still uncertain returns, leaving all firms vulnerable to disappointments in AI payoffs. Model analysis based on such contest motives highlights the downside risk of current AI exuberance. As competitive pressure drives capex higher, the net economic surplus – the total payoff less investment costs – declines for the sector as a whole and could turn negative in adverse scenarios (Graph 11.B). Disappointment in returns could trigger a sudden pullback in financing and turn the capex boom into a protracted investment bust, with potential knock-on effects on financial conditions (see below).
Another risk is that the AI boom runs into a supply side roadblock. The AI build- out has recently been facing growing bottlenecks in electricity, advanced semiconductors and grid equipment. Fast-growing demand for computing power is already pressuring electricity prices and input costs, with potential spillovers to inflation. Looking ahead, these temporary shortages may also amplify over-investment, as firms attempt to lock in future capacity through long-dated contracts that further expose them to any disappointments in demand.
...
Should inflation rise significantly or AI-led investment turn to a bust, the macroeconomic consequences could be amplified by existing financial vulnerabilities. A tightening of policy rates needed to contain inflation could precipitate a sharp pullback in asset prices after a prolonged period of exuberant risk-taking, triggering disruptive macro-financial feedback loops. A reversal of AI optimism could likewise have major financial consequences, given AI firms’ rising leverage and growing footprint in credit markets. Vulnerabilities extend to their supplier ecosystem, including engineering, procurement and construction (EPC) contractors whose balance sheets are comparatively weak, leaving them exposed to any capex pullback by hyperscalers.
...
A sharp repricing of equity risk could prompt a reassessment of corporate credit risk and lead to tighter credit conditions more broadly.1 Indeed, broad indices of credit spreads tend to correlate negatively with stock market returns (Graph 14.A), more so for the high-yield than the investment grade segment. While large, synchronised corrections in both markets are rare, there are notable precedents such as the Great Financial Crisis and the March 2020 dash for cash episode. A repricing of risk this time, whether triggered by higher interest rates or an AI bust, has the potential to be similarly disruptive by triggering a corporate credit freeze with wider implications for aggregate investment."
There’s no question we’re in a massive AI bubble, the only question is how do we get out of it without wiping out the broader economy.
I think it’ll be the printing press, not least because of moonshot executive pay tied to $price
It's ether going to be one catastrophic crash or decades of stagnation and decline.
living standards, but not asset prices
Pretty much. Given the levels of investment in capital and research if AI companies actually hit what they're aiming at they'd have to collapse the labor market to recoup, bricking the economy in the process. Given the levels of outside investment inflating valuations, if the bubble pops it's 2008 all over again. There's this incredibly narrow window of "just useful enough to extract rents" where everything doesn't go to shit.
Extract rents? I don’t think you’re using the term rent correctly here.
No, it's correct. The best (short-term) case is that they become eternal parasites. If they fail to do that, they'll bring a lot down with them when they fall.
>The best (short-term) case is that they become eternal parasites.
Producing a product that delivers value and people are willing to pay for makes you a "parasite"? Sure, it might cause massive disruptions to the labor market, but that's mostly orthogonal to whether it's a "parasite" or not. Mechanized farming has almost wiped out agricultural employment (compared to pre-industrial levels), but that doesn't make tractor manufacturers or fertilizer companies "parasites"
tractor manufacturers or fertilizer companies didn't suck down the work of generations of predecessors in a questionably-legal fashion only to turn around and sell a heavily discounted version of that back to them. I'm not sure where "parasite" becomes appropriate, but your analogy is poor.
>in a questionably-legal fashion
Maybe in the eyes of seething artists/programmers seeing their jobs getting automated, but courts have so far ruled that AI training falls under fair use.
Moreover it's not hard to think of vaguely similar objections to fertilizers. They're often produced at some harm to society, as well as their use. They're also in some sense, a "heavily discounted" versions of that they replaced, bird guano or whatever.
re: questionable legality https://arstechnica.com/tech-policy/2025/02/meta-torrented-o...
> Moreover it's not hard to think of vaguely similar objections to fertilizers.
It's completely different. If LLM companies pulled this out of thin air it would be also different, but no; they've effectively plundered the commons and locked up all the profit for themselves. If intellectual labor goes the way of agricultural labor, I think humanity will have lost something valuable.
And don't come back with the "farmers would have said the same thing about the industrial revolution!" thing again if you're just going to terminate your thought there. Automating agricultural labor brings vast material benefits for all since it lowers the cost of tangible goods needed for life. I'd challenge you take this one step further and explain why automating intellectual labor will provide similar fruits and is therefore something to cheer for.
If we actually do meaningfully automate intellectual labor, we create a world where we have real technical solutions for our toughest problems. Maybe we can get carbon capture and fusion energy working. There’s a theoretical world of abundance for us to explore.
That’s the steel man argument.
FWIW I mostly don’t believe that LLMs are the answer, I don’t think they’re going to reach a high enough level of capability to do this, and I think the current AI companies are problematic in a lot of ways.
I also think LLM use is bad for us and probably harms our thinking abilities. And using it takes away a lot of what it means to be human.
Personally I like both physical and mental difficulty. I like gardening even if I could just buy mass produced flowers. I like riding a bike even though cars are “easier”. I like playing ukulele with my family even though I can barely make a chord, much better than listening to some other real musician, or Suno ai generated songs. I like eating my wife’s sourdough bagels even if they take several hours more than just buying some.
And I think having those regular challenges and achievements make life worth living! And I worry that the AI future that some envision will make much of what we get value from feel meaningless in the same way that writing code by hand is starting to.
Maybe we’ll still be fine in the same way I find meaning in all of those things that I listed above. But damn what a gamble
No, that's actually more destructive in the short term. That was the OP's point. If they actually become something effective, they will utterly destroy the economy. If they fail, they will drag the economy down around them as they go. The least destructive possible outcome is becoming eternal parasites.
Your analogy would be less terrible if mechanized farming took the fruits of traditional farmers and repackaged it.
Well there's also the fact that fundamentally and ultimately, AI is incompatible with the economic system. Capitalism is rooted in human labour having positive economic value, and hence demand. AI will ultimately automate all labour, making the economic value 0. Eventually capital generation will simply die and the system crashes.
Meanwhile AI has gotten so good it can just about one shot a SaaS app.
I’m not worried about it…
Yeah but the investments arent aiming for churning out SaaS apps. Its to automate large swathes of intellectual labour. Of which only SWE has been cracked yet. There is a question mark as to if the others will crack. If they arent then these investments will collapse from speculation down to reality. That possibility is what is being discussed here
As to whether that will happen, I think that risk is real. Because claude code isnt made by the generalozed capabilities of the tech but by good old non-generalozable hueristics and rule based engines. I dont think that will scale to other feilds at the factor these investments assume. Its the bitter lesson again. It scales with deliberate and specific design, not data, so it wont scale
We learnt this with ibm watson. Deepblue achieved chess supremacy but the last mile wasnt data driven, it was heiristic driven, and so watson, its successor, couldnt scale/generalize.
My prediction is that this speculation on LLMs with harnesses will collapse since they wont scale. We'll have another winter where the reasearchers will be leaft alone long wnough to come up with the next breakthrough (probably game theory based data driven agency) which might then create what this hypecycle is speculating
There’s argument to be made that SWE hasn’t been cracked either. The latest models are great at coding medium sized applications, but figuring out the requirements and consolidation of domain knowledge is something still lacking
Yeah. I think LLMs wont be able to do that on their own or with hueristics. We'll need to bake game theory into the base model for that
you mentioned a very good point about scalability. we're seeing alot of productivity gains, but only from SWEs, which are but a very small segment of the global economy. all other economic use cases require thorough last-mile development and iteration that is not too different with current automation tools.
A friend who is a psychologist was telling me he thinks in another year or two insurance companies will insist people see an AI therapist first before being willing to pay for a real person.
LOL, the same AI that has landed companies in court defending themselves against wrongful death lawsuits for helping someone convince themselves suicide is the right answer, and even encouraging them? That AI? I am unclear that any insurance company is going to want a piece of that action anytime soon.
What you've just told me is that psychologists, just like SWEs, are prone to thinking they know how business works but in fact know fuck all.
Tracks. This is that same speculation that AI will be good enough. Wonder what a crash in that isecase will look like? Increased suicide rates? More instances of psychosis? It might not wven be directly measurable or easily traceable to AI therapists. Would suck
All those automaton tools will eventually be initially one-shotted and then monitored by LLMs though. There probably won't be a "last mile" per se; just constant tweaking and optimizations throughout, within a feedback loop.
What your describing is iterating in the last mile. Your assuming that the AI will iterate in the last mile with the same efficacy that it iterates before that. I think that will fail. Thats the bitter lesson, that adhoc solutions to the last mile (rather than generalozed solutions that scale with training data) asymtotally stall and so dont scale.
Meaning claude code wont be able to make a "claude video editing" or "claude accounting" with the current tech. Human experts will need to encode their knowledge into it for the last mile and that wont scale the way these speculations expect
We aren't seeing productivity gains in software either. What we are seeing is a lot of people who claim to be more productive, but in fact are building piles of tech debt that will fall over before long. But hey, they're building that tech debt really fast!
> but in fact are building piles of tech debt that will fall over before long
This is speculation as well. Its well founded but speculation nonetheless. Youre speculating things will stay the way they have till now.
I do see your point but what makes me consoder the other side is that ive been building an app that reaches ~10k LOC, purely with opus, no code review at all, and it hasnt hit any tech debt issues that i havent easily been able to address. Setting up good context management meant that claude could just figure things out itself.
And for reference this is an app that manages an ethernet camera, runs vehicle detection on the stream, and surfaces the detections on an ipad for operators to inspect and annotate for cellphone usage, so not trivial. Needed good architechtong and design from my end, but it was honestly scarily easy. So idk what the threshold for tech debt crash is but it wasnt there
We will find out how much of work is given to people just so that there's a person/company associated with a technical decision. I personally think this might be quite high.
we already know this, the term bullshit jobs exist for this reason.
Exactly. I build automation tools for my company which have improved productivity quite a lot and put precisely zero people out of a job. Partly we find other things for them to spend time on, and partly it just turns out that we like to have humans doing jobs.
It is cliche at this point that HN is the place you go to hear software developers reduce all of the world's problems into simple algorithmic arguments which for some reason never actually solve anything. Not shocked that we are similarly incapable of understanding that algorithmically replacing a software developer isn't easy just because we think we know what the job is.
> Of which only SWE has been cracked yet.
When did that happen?
While I don't agree with your premise at all, even if it could one shot a SaaS product (a statement so vague it's meaningless) I don't think there's much of an argument for why that's economically useful. A lot of SaaS has free software/open source equivalents anyway (how else do you think the clanker's are able to plagiarize it?). People still pay for Office even though you could easily use LibreOffice, or GitHub when you could self host Forgejo. It's like when anthropic made a big deal out of making a broken compiler. Neat, so, after ingesting all of open source and burning a trillion tokens you ended up with something worse than what's already out there; and instead of doing something economically useful like giving a person money to build it or supporting the open source ecosystem, you're just wasting energy on datacenters.
Why even build SaaS apps? AI can just do what a SaaS app can
So good at it that I’m right now in the process of building instead of buying.
Here’s how that plays out in the economy:
- My company spent $50 on my tokens to build this internal tool
- Anthropic spent $XXX to deliver those tokens to me.
- The company I was going to buy the tool from lost $XX,XXX per year that I would have paid them.
I dunno, kind of sounds like the economy just got smaller.
I could usually accept the idea that software getting cheaper generally increases demand for software and expands the economy surrounding it, but I’m not sure if we have precedent for what happens when software becomes positively worthless.
Are you putting the $XX,XXX-50 under your mattress or investing it in something else productive?
I certainly can’t give you a better answer for my company than “it depends” or “I don’t really know.”
The company could just be happy to have better margins and be happy the stock finally went up. It might literally do nothing with them or do something economically unproductive like buy back stock.
What I can tell you with certainty is that we aren’t going to hire anyone else or launch any other product as a result. Our business just isn’t at that level of growth potential.
Perhaps we can surmise that money going to shareholders can grow the economy. They’ve got more money to reinvest in other stuff.
But then again, if everyone can shart out a SaaS app with $50 in tokens, what software companies will they want to invest in?
AI gives me that feeling of “what happens to bakers and butchers when the supermarket gets invented and they decide to sell bread and meat at or below cost?”
This is the right question.
Every company has a list of >WACC IRR projects that it can spend saved money on. If not, it’s a cash cow company that wasn’t growing in the first place and will allow shareholders to use the saved cash for other economically expanding projects.
What companies can expand if the income of consumers is shrinking. This is the scary bit to me — AI crashes and takes the economy with it, or; AI succeeds as promised and people go unemployed and crash the economy.
The only path that isn’t disastrous is threading the needle of “just right” productivity gains. The people in charge aren’t smart enough to give me warm fuzzy feelings on that.
What software companies will economically expand if the price ceiling on software is really low?
Economy != software companies. Maybe there needs to be a capital rotation out of tech? That’s speaking beyond my expertise though (and imo is a little too doomer). Continuing the hypothetical through: Healthcare, industrials, financial services and many other verticals have plenty of growth opportunities.
Otherwise it would probably be the software companies that are the most focused on last-mile details (where AI in my experience has the most trouble). I expect that as consumers are faced with more and more AI slop SaaS they will be increasingly willing and able to pay for quality.
Unless that money is being spent on more tokens, it'll probably be used for stock buybacks.
Which enrich existing shareholders who can use that capital to invest in other economically expanding projects.
And if the company didn't need $XX,XXX*0.90 (or more) that you would have paid them to further develop their product and stay in business? If that other company now paid their own $50 in tokens? Maybe the overall flow of money in the economy went from $XX,XXX (you) + $XX,XXX (them) + $(not much, AI didn't exist yet) equals or is greater to $50 + $50 + ($xbillions in AI)? Dunno.
that is not production ready
Eh, that's not been my company's experience. Error reports are down. Performance is up clients are happy.
Pretty soon we're going to have to reckon with the fact that AI writes better code than us.
so your company runs on a vibe-coded saas app, that sure is a confidence booster for your would-be customers.
How will the customers know? And if it does what they need it to do, then why would they even care?
wouldn't take long for them to find out with the developer bragging about it. he one-shot vibe-coded a sass app (keyword: one-shot), that says alot about the quality.
Sounds like they’re happy.
And own nothing?
Anything we can see?
They never want to show it
What does that have to do with the article?
> Meanwhile AI has gotten so good it can just about one shot a SaaS app.
There isn't a direct correlation between AI improvement or stagnation and whether or not the amount being spent by AI labs and the associated ecosystem will result in a financial crash.
Look into the history of railroads and the internet itself to see how massive levels of investment can result in economic crashes even when the thing being invested in produces real, widespread societal value.
One could argue that one of the nightmare economic scenarios for AI is actually that it gets too good too fast and results in a wipeout of the white collar worker that we are currently nowhere near ready to deal with given how propped up our economy is on consumer spending.
difference this time is they have "fiat money" and money printer. Market and all inv. bankers knows that in major crash they will print unlimited amounts so back to same prices or near them. printer is still printing and it's only goes to selected investments
not at current inflation levels, no way to "print" if it means causing inflation to spike beyond unhealthy levels.
As if they did not back then. Fiat is just simpler to work with, but one can pull a bubble without it just fine. Anyone forgot the railroad crash of 1873? The tulip mania of 17th century?
The dot com bubble of 2000
The Nasdaq took 14 years to recover, 17 once you factor in inflation.