I would add more metrics to think about. For example, very few people used Internet in the dotcom era while now the AI use is distributed into all the population using the Internet that will probably not growth too much. In this case, if Internet population is the driver, and it will not growth significantly we are redistributing the attention. Assuming "all" society will be more productive we will all be in the same train at the relatively same speed.

The 90s bubble also had massive financial fraud and laid capital that wasn’t used at 100% utilization when it hit the ground like what we are seeing now.

It’s different enough that it probably isn’t relevant.

Cheaper cloud costs

Video game crash followed by video games taking off and eclipsing most other forms of digital entertainment.

Dot com crash followed by the web getting pretty popular and a bit central to business.

To all those betting big on AI before the crash:

Careful, Icarus.

The biggest problem is the infrastructure left behind from the Dotcom boom that laid the path for the current world (the high speed fiber) doesn't translate to computer chips. Are you still using intel chips from 1998? And the chips are such a huge cost, and being backed by debt but they depreciate in value exponentially. It's not the same because so much of the current debt fueled spending is on an asset that has very short shelf life. I think AI will be huge, I don't doubt the endgame once it matures. But the bubble now, spending huge amounts on these data centers using debt without a path to profitability (and inordinate spending on these chips) is dangerous. You can think AI will be huge and see how dangerous the current manifestation of the bubble is. A lot of people will get hurt very very badly. This is going to maim the economy in a generational way.

Dotcom mania companies were not Internet providers. They tried making money on the internet, something people already saw as worth paying for.

And what were the societal benefits of the internet?

The analogy to the dot com bubble is leaky at best. AI will hit a point of exponential improvement, we are already in the outer parts of this loop.

It will become so valuable so fast we struggle to comprehend it.

> [At dotcom time] There was a bubble, many non-viable companies got funded and died, and nevertheless the internet did eventually change everything.

It did, but not for the better. Quality of life and standard of living both declined while income inequality skyrocketed and that period of time is now known as The Great Divergence.

> He's (unsurprisingly) making an analogy to the dotcom bubble, which seems to me correct.

He's got no downside if he's wrong or doesn't deliver, he's promising an analogy to selling you a brand new bridge in exchange for taking half of your money... and you're ecstatic about it.

My team is doing the same, and yet all of us still aren't sure that we're actually more productive overall.

If anything it seems to me like we've just swapped coding with what is effectively a lot more code review (of whatever the LLM spits out), at the cost of also losing that long term understanding of a block of code that actually comes from writing it yourself (let's not pretend that a reviewer has the same depth of understanding of a piece of code as an author).

Many AI startups around LLMs are going to crash and burn.

This is because many people have mistaken LLMs for AI, when they’re just a small subset of the technology - and this has driven myopic focus in a lot of development, and has lead to naive investors placing bets on golden dog turds.

I disagree on AI as a whole, however - as unlike previous technologies this one can self-ratchet and bootstrap. ML designed chips, ML designed models, and around you go until god pops out the exit chute.

> This is going to maim the economy in a generational way.

Just as I'm getting to the point where I can see retirement coming from off in the distance. Ugh.

> the current debt fueled spending

Honestly I think the most surprising thing about this latest investment boom has been how little debt there is. VC spending and big tech's deep pockets keep banks from being too tangled in all of this, so the fallout will be much more gentle imo.

One thing to note about modern social media is that the most negative comment tends to become the most upvoted.

You can see that all across this discussion.

Companies want GW data centers, which are a new thing that will last decades, even if GPUs are consumable and have high failure rates. Also, depending on how far it takes us, it could upgrade the electric grid, make electricity cheaper.

And there will also be software infrastructure which could be durable. There will be improvements to software tooling and the ecosystem. We will have enormous pre-trained foundation models. These model weight artifacts could be copied for free, distilled, or fine tuned for a fraction of the cost.

Cisco, Level3 and WorldCom all saw astronomical valuation spikes during the dotcom bubble and all three saw their stock prices and actual business prospects collapse in the aftermath of it.

Perhaps the most famous implosion of all was AOL who merged (sort of) with TimeWarner gaining the lion's share of control through market cap balancing. AOL fell so destructively that it nearly wiped out all the value of the actual hard assets that TW controlled pre-merger.

The problem is the hype machine that is claiming it will replace human labor which justifies the insane losses.

These companies are burning cash to support the current formulation of AI services.

These services will survive because they are useful but probably not at its current cost

A real industry can be a bubble too. Not every bubble looks like tulips.

For example:

- Dotcom bubble. Of course making website was and is a real industry.

- Japanese real estate bubble. Of course building houses was and still is a real industry. It's so real people call it real estate, right.

irony level IV

Apparently not. An LLM proofread would've definitely caught that

There aren't very many IPOs in general. There were about 8000 publicly traded companies in the US in Jan 2000. Today there are about 3950. A lot of the AI related IPOs have been the infra like CoreWeave and Nebius.

there is no way they could raise that much money from public markets.

Also, there s no need to invent new tech names anymore. Marketing can add "AI" to the company name, or (as they say), change the wording from "Loading..." to "Thinking.."

Many argument the current batch of models provide a large capability overhang. That is, we are still learning how to get the most out of these models in various applications

Bezos didn't define "society", but knowing Devil is what Devil does, we can infer:

1. Amazon files the most petitions for H1-B work visas after Indian IT shops. 2. Amazon opposed minimum wage increase to $15/hr until 2018! 3. Amazon not only fires union organizers, it's claiming National Labor Relations Board is unconstitutional!

Once upon a time, society was all of us, but Society were the filks that held coming out parties and gossiped about whose J-class yacht was likely to defend the America's cup.

Society with a capital S are the beneficiaries of the bubble.

When billionaires talk about “benefits”, I’m getting very suspicious.

Bad comparison.

The leap of faith necessary in LLMs to achieve the same feat is so large its very difficult to imagine it happening. Particularly due to the well known constraints on what the technology is capable of.

The whole investment thesis of LLMs is that it will be able to a) be intelligent b) produce new knowledge. If those two things that dont happen, what has been delivered is not commensurate to the risk in regards to the money invested.

And a lot of the gains from the Dotcom boom are being paid back in negative value for the average person at this point. We have automated systems that waste our time when we need support, product features that should have a one-time-cost being turned into subscriptions, a complete usurping of the ability to distribute software or build compatible replacements, etc..

The Dotcom boom was probably good for everyone in some way, but it was much, much better for the extremely wealthy people that have gained control of everything.

> Careful, Icarus.

What does that even mean?

pets.com was a fat loser only telling telling people that were going to fly.

Amazon was Icarus, they did something.

Vs weak commentators going on about the wax melting from their parents root cellar while Icarus was soaring.

Most of Y Combinator are not using AI they just say that and you're worried about the people who do things?

There are probably a lot of cool and useful things you could do with a bunch of data centers full of GPUs.

- better weather forecasts

- modeling intermittent generation on the grid to get more solar online

- drug discovery

- economic modeling

- low cost streaming games

- simulation of all types

That's true for the GPUs themselves, but the data centers with their electricity infrastructure and cooling and suchlike won't become obsolete nearly as quickly.

If you look at year over year chip improvements in 2025 vs 1998, it's clear that modern hardware just has a longer shelf life than it used to. The difficulties in getting more performance for the same power expenditure are just very different than back in the day.

There's still depreciation, but it's not the same. Also look at other forms of hardware, like RAM, and the bonus electrical capacity being built.

We don’t have moores law anymore. Why are the chips obseleting so quickly?

This is not really true, e.g. Cogent was basically created by buying bankrupt dotcum-bubble network providers for cents on the dollar.

Those companies are also likely still in the red. They're banking on the hope that one day they will be profitable. I'm sure one of them will be.

> 3 different AI services

Why three? Will you ever be in a position where one will do it for you?

> and most of my team is also paying money for various AI stuff.

And what are they using it for?

> Sounds like a real industry to me.

Sounds like early adopter syndrome to me. We'd have to know more about your business to take this out of the realm of hazy anecdotes.

are you paying them enough to have profitable unit economics? How hard would it be for you to switch from one provider to another?

Typos are proof something wasn't Ai written. Everyone should make at least one tyop in their writing.

Then it’s a market for lemons. Buyers can’t (or choose not to) tell a good AI idea from a bad one.

IPOs aren't what they once were. The burden of being a public company has increased (SOX and related public company costs are $5-10M/year), so companies are far more likely to stay private. That has created a positive feedback cycle as the private funding ecosystem has become increasingly robust, which is why you see so many $100B+ private companies.

Also keep in mind that the biggest companies during that bubble had peak market caps of ~500B and then lost ~90%, so 400-500B in losses each and total internet related losses of a couple trillion. If NVDA lost 90%, it would be down 4 trillion dollars, or twice that total just by itself.

AI company valuations collapsing would have meaningful impacts on the broader market. Big pension/mutual funds are important sources of capital across every sector, and if they're taking big losses on NVDA, GOOG, and a portfolio of privates, it will have a chilling effect on their other activity.

This time is always different, until it isn't.

However, it is different from the internet bubble partially for the reason you describe.

There have been a few IPOs, but they perhaps happened earlier in the cycle, or companies are pivoting into AI. I'm thinking companies like Palantir, which was always AI, or Salesforce which is making a big AI pivot.

Most of the funding is not coming from public sectors. There is so much private capital available that it isn't necessary. I believe the bubble is in VC, which some would think is find because it protects public markets from the crash, but I'm not sure that is correct.

When the VC money stops flowing into AI, I think it will send a shockwave through the public markets. The huge valuations of companies like OpenAI, Anthropic, etc will be repriced, which will probably force a re-pricing of public darlings like Palantir, Microsoft, NVIDIA.

If VC funds aren't buying NVIDIA chips and building data centers, everyone will feel the need to re-price.

It's emotional, not logical.

Why do you think the models will become out of date and less useful? Like, compared to what? What external factor makes the models less useful?

If it's just to catch up with newly discovered knowledge or information then that's not the model, they can just train again with an updated dataset and probably not need to train from scratch.

If they stop getting returns in intelligence, they will switch to returns in efficiency and focus on integration, with new models being trained for new data cutoffs if nothing else. Even today there is at least a 5-10 year integration/adoption period if everything halts tomorrow.

There is no reality in which LLMs go away (shy of being replaced).

Models don't become out of date now that deep research exists.

It will benefit all people but it will disproportionally benefit more rich people.

Manna ponders the same question. I don't think we're headed towards the "good" ending.

https://marshallbrain.com/manna1

Which of those did the Internet benefit?

It is all society as long as they have access, and they do. Even if the big labs get more closed off, open source is right there and won’t die.

AI increases everyone’s knowledge and ultimately productivity. It’s on every person the learn to leverage it. The dynamics don’t need to change, we just move faster and smarter

Counter prediction. AI is going to reduce the (relative) wealth of the tech companies.

AWS and Facebook have extremely low running costs per VPS or Ad sold. That IMO is one of the major reasons tech has received its enormously high valuation.

There is nuance to that, but average investors are dumb and don't care.

Add in a relatively high fixed-cost commodity into the accounting, and intuitively the pitch of "global market domination at ever lower costs" will be a much harder sell. Especially if there is a bubble pop that hurts them.

What value does your question add, if it wasn't Bezos saying this would you have the same question?

How are you defining rich, Billionaires? It's sad that your comment is the top post.

About 40% of AI infrastructure spending is the physical datacenter itself and the associated energy production. 60% is the chips.

That 40% has a very long shelf life.

Unfortunately, the energy component is almost entirely fossil fuels, so the global warming impact is pretty significant.

At this point, geoengineering is the only thing that can earn us a bit of time to figure...idk, something out, and we can only hope the oceans don't acidify too much in the meantime.

That everybody all over the world can instantly connect with each other?

Yes there is a very real trade off between labour and capital.

In the past the tradeoff has been very straight forward. But this is a unique situation because it involves knowledge and not just the physicality of the human in regards to productivity.

> Why three? Will you ever be in a position where one will do it for you?

I believe LLMs will be niche tools like databases, you pay for the product not 'gpt' vs 'claude'. You choose the right tool for the job.

I have a feeling coding tool with be separate a niche like Cursor, which LLM it uses doesn't matter. It's the integration, guard rails, prompt seeding, and general software stuff like autocomplete and managing long todos.

Then I pay for ChatGPT because that's my "personal" chat LLM that knows me and knows how I like curt short responses for my dumb questions.

Finally I pay for https://www.warp.dev/terminal as a terminal which replaced Kitty terminal on macos (don't use it for coding) which is another niche. Cursor could enter that arena but VSCode terminal is kinda limited for day-to-day stuff given it's hidden in an larger IDE. Maybe a pure CLI tool will do both better.

there's some lock-in for both chatgpt (the history and natural chat personalization feature is super useful) and with Cursor I'm fully invested in the IDE experience.

The lie is that LLMs are the product itself rather than the endless integration opportunities via APIs and online services.

The costs are a weak argument. The more stronger argument for why they arent going public any time soon is that OAI in particular is a corporate governance nightmare, in which the way they transmit information about their firm and financials will have to completely change.

Theres also plenty of money washing around in private markets so no need to go public. Staying private is an advantage.

I see what ya did tbere :)

Historically, the worst busts following the bursting of an asset price bubble, in terms of real economic impact, have been from debt fueled bubbles (Great Depression, Global Financial Crisis). You can read Hyman Minsky and Irving Fisher for a detailed analysis of why, but it mainly comes down to the fact that the financial obligations remain once prices and expectations have reset.

Then you have the busts that follow public equity fueled bubbles (Dotcom crash). Nowhere near as bad as the former, but still a moderate impact on the economy due to the widely dispersed nature of the equity holdings and the resulting wealth effect.

What we have now is more of a narrowly held private equity bubble (acknowledging that there's still an impact through the SP500 given widespread index investing). If OpenAI, Anthropic, Perplexity, and a bunch of AI startups go bust, who loses money and what impact does it have on the rest of the economy?

> What external factor makes the models less useful?

Life. A great example can be seen in the AI-generated baseball-related news articles that involve the Athletics organization. AI articles this year have been generating articles that incorrectly state that the Atlanta Braves played in games that were actually played by the Athletics, and the reason is due to the outdated training model. For the last 60 years before 2025, the Athletics played in Oakland, and during that time their acronym was OAK. In 2025, they left Oakland for Sacramento, and changed their acronym to ATH. The problem is that AI models are trained on 60 years of data where 1. team acronyms are always based on the city, rather than the mascot of the team, and 2. acronyms OAK = Athletics, ATL = Atlanta Braves, and ATH = nothing. As a result, an AI model that doesnt have context "OAK == ATH in the 2025 season" will see ATH in the input data, associates ATH with nothing in it's model, and will then erroneously assume ATH is a typo for ATL.

How was Palantir always AI?

Is adding new data to a model a full retraining from scratch? Or can it be added on top of an existing model?

If it costs $10B to add 1 year of data to an existing model, every year, that doesn’t sound too good.

> If they stop getting returns in intelligence, they will switch to returns in efficiency

I don't think we can assume that people producing what appear to be addictive services are going to do that, especially when they seem to be addicted themselves.

So with every prompt you are expected to wait that long? I highly doubt general people will be willing to wait, it also doesn't seem entirely viable if you want to do it locally, less bandwidth and no caching akin to what a literal search engine of data can do.

They probably are getting a bonanza of intelligence from people using them.

> It will benefit all people but it will disproportionally benefit more rich people

The rich already have a diminishing returns situation with money. Everyone else has much more upswing.

The wealth gap widening is quite independent from AI being involved. A natural progression which was always happening and continues to be happening. Entil some sort of catastrophe reshuffles the cards. Usually a war or revolution. The poor simply rising up or a lazy and corrupt ruling class depriving their country of enough resources and will to defend itself that some outside power can take it.

If be benefitting you mean displacing all reason to live, then yes, it will solve that problem I face. Now I will be certain.

What an incredible read - thanks for the share

The fact that Bezos is saying this is precisely why the commenter is asking this. He clearly stands to benefit massively from the bubble. Statements like this are meant to encourage buy-in from others to maximize his exit. Presumably "rich" refers to those, like Bezos, who already have incredibly disproportionate wealth and power compared to the majority of people in the US. I'm honestly not sure what the thrust of your comment even is.

History would indicate that if the you make a bubble big enough and then it pops, you don't have to clean up the mess. 2008 wasn't that long ago.

Yeah, but net worth is weird because for most people, it just measures age. When you're young, you have nothing in your 401k and you have a brand new mortgage, so you're worth around $0. Negative if you have any student loans.

When you're in your 50s or 60s, the mortgage is repaid, and if nothing blew up, you probably also have a million or two in your 401k, so at that point, it's actually not that hard for a person who had a decent career in the SF Bay Area to be worth $4M+. And many FAANG retirees will probably flirt with $10M+ if they don't spend too much.

Interesting. Do you have any sources for this 60/40 split? And while I agree that the infrastructure has a long shelf life, it seems to me like an AI bubble burst would greatly depreciate the value of this infrastructure as the demand for it plummets, no?

It's extremely regional. I was at 160 after less than 5 years in the industry. In the Bay Area, that still put me in the bottom 10% for tech workers

https://www.levels.fyi/heatmap/

Midway through my career, these days I feel fortunate to be making 60k. But I'm in Canada.

Right, so not a bubble then

Even then, the people the money came from voluntarily chose those institutions. If you don't like risky investments, there are lower risk institutions or products you can put your money in. It's still ultimately the will of the individual what they do with their money, and the consequences of bad choices are still mostly contained to the individuals who make them. But when it's done with tax money, everyone's dragged into it and has no personal choice in the matter. Even worse, when it's tax money without democracy, even if the majority of people don't like how it's used, they even collectively have no choice in the matter.

You made the point for me. That 100bn doubled every 2-3 years. It wasn’t a bubble, but it absolutely looked like one. This will be a bitter lesson too.

No clue

Then why has my experience with AI started to see such dramatically diminishing returns?

2022-2023 AI changed enough to be me to convert from skeptic, to a believer. I started working as an AI Engineer and wanted to be on the front lines.

2023-2024 Again, major changes, especially as far as coding goes. I started building very promising prototypes for companies, was able to build a laundry list of projects that were just boring to write.

2024-2025 My day to day usage has decreased. The models seem better at fact finding but worse for code. None of those "cool" prototypes from myself or anyone else I knew seemed to be able to become more than just that. Many of the cool companies I started learning about in 2022 started to reduce staff and are running into financial troubles.

The only area where I've been impressed is the relatively niche improvements in open source text/image to video models. It's wild that you can make sure animated films on a home computer now.

But even there I'm seeing no signs of "exponential improvement".

While this remains possible my main impression now is that progress seems to be slowing down rather than accelerating.

Very few people predicted LLMs, yet lots of people are now very certain they know what the future of AI holds. I have no idea why so many people have so much faith in their ability to predict the future of technology, when the evidence that they can't is so clear.

It's certainly possible that AI will improve this way, but I'd wager it's extremely unlikely. My sense is that what people are calling AI will later be recognized as obviously steroidal statistical models that could do little else than remix and regurgitate in convincing ways. I guess time will tell which of us is correct.

Thank you for acknowledging this. The internet was created around a lot of lofty idealism and none of that has been realized other than opening up the world's information to a great many. It made society and the global economy worse (occidental west; Chinese middle class might disagree) and has paralleled the destabilization of geopolitics. I am not luddite but until we can, "get our moral shit together" new technologies aren't but fuel on the proverbial fire.

If you work in a team then you are likely already not writing most of the code yourself.

There will be point where ai will consistently write better prs - you can already start to see it here and there - finding and fixing bugs in existing code, refactoring, writing tests, writing and updating documentation and prototyping are some examples of areas where it often surpasses human contribution.

HN said the same thing about Uber forever.

"Claude, make sure to sprinkle at least one typo in here. And no em dashes you hear?"

Only worms until the typos get into the training data. You need to say at least one thing no one has ever said before in every post.

A market for lemons means there is an information asymmetry. Sellers know what they have and try to offload their lemons on clueless buyers. I don't think that's the case here.

It may be true that OAI et al are raising money in private markets, but does that matter? Ultimately they are still just raising money. Ultimately returns need to show up. You cannot escape that. If you cannot do that nobody will eventually supply the funds to keep operating.

The big advantage of staying private is controlling the narrative.

Not necessarily. The benefits may barely trickle down, and the conditions for the majority could degrade overall.

> It will benefit all people but it will disproportionally benefit more rich people.

Yes: you'll be homeless and living under a bridge, but you'll have an LLM therapist on your phone to console you. That's a benefit!

What you mean is those with some form of ownership of the technology. If development eventually results in full automation, with the expense of production reduced to zero, money will be irrelevant.

Rich people will enjoy additional monetary benefits, but everyone will still enjoy the same standard benefits.

It will not benefit the rich as disproportionately as the covid pandemic did.

And we're not talking a 51-49 disproportion, we're talking about a 99.9999 percent of the benefit will go to the richest people

Would you rather be a 22 year old starting in life in 2025 or 1995? Unless you pick one of the few countries that underwent a drastic change of regime in that time, the answer’s pretty clear to me.

Is the average person actually better off after the late 90’s internet is probably a harder question than it might seem.

The long tail may be closer to what I want, but the quality is also generally lower. YouTube just doesn’t support a team of talented writers, Amazon is mostly filled with junk, etc.

Social media and gig work is a mixed bag. Junk e-mail etc may not be a big deal, but those kinds of downsides do erode the net benefit.

> AI increases everyone’s knowledge and ultimately productivity. It’s on every person the learn to leverage it. The dynamics don’t need to change, we just move faster and smarter

This is incomplete in key ways: it only increases knowledge if people practice information literacy and validate AI claims, which we know is an unevenly-distributed skill. Similarly, by making it easier to create disinformation and pollute public sources of information, it can make people less knowledgeable at the same time they believe they are more informed. Neither of those problems are new, of course, but they’re moving from artisanal to industrial scale.

Another area where this is begging questions is around resource allocation. The best AI models and integrations cost money and the ability to leverage them requires you to have an opportunity to acquire skills and use them to make a living. The more successfully businesses are able to remove or deprofessionalize jobs, the smaller the pool will be of people who can afford to build skills, compete with those businesses, or contribute to open source software. Twenty years ago, professional translators made a modest white collar income; when AI ate those jobs, the workers didn’t “learn to leverage” AI, they had to find new jobs in different fields and anyone who didn’t have the financial reserves to do that might’ve ended up in a retail job questioning whether it’s even possible to re-enter the professional class. That’s great for people like Bezos until nobody can afford to buy things, but it’s worse for society since it accelerates the process of centralizing money and power.

Open source in particular seems likely to struggle here: with programmers facing financial downturns, fewer people have time to contribute and if AI is being trained on your code, you’re increasingly going to ask whether it’s in your best interests to literally train your replacement.

Is AI at the same level of global instant comunication and gigabites of bandwidth?

Entry-level total comp for SWEs in the SF Bay Area is probably $250k. A senior dev at a public tech company can easily clear $400k. Really senior engineers at FAANG routinely clear $1M.

Is that in the world? Or in the USA. Idk I thought there would be more millionaires.

In my experience they invariably conflate LLMs with AI and can’t/won’t have the difference explained.

This is the blind spot that will cause many to lose their shirts, and is also why people are wrong about AI being a bubble. LLMs are a bubble within an overall healthy growth market.

To be clear when I said “finance folks” I wasn’t really referring to VCs. I’m talking more family office types that manage big pools of money that you don’t know about. The super wealthy class that has literally has more money than the King but would be horrified if you knew their name. Old money types. They’re well aware of the “dumb VC” vibe that just throws money after hype. The finance folks I’m talking about are the type that eat failed VCs for lunch.

> More profit for companies

Agreed. So I don't think it's a bubble.

Will also be good for consumers in the long-term: much faster pace of drug discovery and new tech generally.

Given they're referencing Icarus, they seem to agree with you.

Past bubbles leaving behind something of value is indeed no guarantee the current bubble will do so. For as many times as people post "but dotcom produced Amazon" to HN, people had posted that exact argument about the Blockchain, the NFT, or the "Metaverse" bubbles.

AI itself is a manifestation of that too, a huge time waster for a lot of people. Getting randomly generated wrong but sounding right information is very frustrating. Start asking AI questions you already know the answer too and the issues can become very obvious.

I know HN and most younger people or people with otherwise political leanings always push narratives pointing at rich people bad but I feel a lot of tech has made our lives easier and better. It's also made it more complicated and worse in some ways. That effect has applied to everyone.

In poor countries, they may not have access to clean running water but it's almost guaranteed they have cell phones. We saw that in a documentary recently. What's good about that? They use cell phones not only to stay in touch but to carry out small business and personal sales. Something that wouldn't have been possible before the Internet age.

> commentators going on about the wax melting from their parents root cellar while Icarus was soaring.

Icarus drowned in the sea.

Even if you want to put the world into only two lumps of cellar dwellers and Icaruses it is still a group of living people on one side and a floating/semi-submerged pile of dead bodies that are literally only remembered for how stupid their deaths were on the other.

How much more of centralized data center capacity we actually need outside AI? And how much more we would need if we used slightly more time on doing things more efficiently?

They kind of aren't. If you actually look at "how many dollars am I spending per month on electricity", there's a good chance it's not worth upgrading even if your computer is 10 years old.

Of course this does make some moderate assumptions that it was a solid build in the first place, not a flimsy laptop, not artificially made obsolete/slow, etc. Even then, "install an SSD" and "install more RAM" is most of everything.

Of course, if you are a developer you should avoid doing these things so you won't get encouraged to write crappy programs.

FLOP/s/$ is still increasing exponentially, even if the specific components don't match Moore's original phrasing.

Markets for electronics have momentum, and estimating that momentum is how chip producers plan for investment in manufacturing capacity, and how chip consumers plan for deprecation.

If you keep up with the research this isn't the case, ML timelines have always been slower than anyone likes

Maybe once we get another architecture breakthrough, it won't feel so slow.

I'm not so sure about this.

First were the models. Then the APIs. Then the cost efficiencies. Right now the tooling and automated workflows. Next will be a frantic effort to "AI-Everything". A lot of things won't make the cut, but absolutely many tasks, whole jobs, and perhaps entire subsets of industries will flip over.

For example you might say no AI can write a completely tested, secure, fully functional mobile app with one prompt (yet). But look at the advancements in Cline, Claude code, MCPs, code execution environments, and other tooling in just the last 6 months.

The whole monkeys typewriters shakespeare thing starts to become viable.

Glad to be in agreement. The higher message here is that technology is no substitute for politics, cue crypto-hype which produced little more than crime and money-laundering. Without proper policies, corruption invades every strata of society.

All the comments of AI writing code and making PRs remind me a lot of all the promises about self driving cars. That was more than 10 years ago and today I still don't know anybody that has a car that drives itself. Will AI write useful PRs one day? Probably. Will it do that before I retire or die of old age? Considering I have been using agents for about a year or so, and seen little to no improvement in that time, I'm afraid the current version of AI probably has already peaked and we'll only see marginal improvement due to diminishing returns.

For almost 20 year, Amazon has been the poster child of "A company can be unprofitable for years and still turn out a winner", but of course - not all companies can pivot from being a regular e-commerce company to cloud infrastructure/hosting, and become a money machine.

So the question, at least to me, is how these AI companies will find a product or service that makes them profitable. Other than becoming actual monopolies in their current domains.

R9k wins again

Because the comment was specifically pointing out that this doesn't seem like a bubble because there aren't many IPOs or public pure play AI companies.

It is indeed hard to tell how many of the people selling this stuff are True Believers. It's also a bit scary, given how incredibly implausible some of these stuff they're saying is.

Yeah it’s a complicated picture and of course nobody knows, but it would be helpful to split “benefits” into things like;

- net benefits to the average person (considering drawbacks)

- overall relative benefits compared to income groups

- benefits in certain areas of society and topics

I think there’ll be some “benefits for all” in terms of things like medical advances and health technology. There will also be broader benefits to all in general areas but as a parent poster said it’ll benefit equity holders most and there might be some bad tradeoffs (like we’ll have access to much better information and entertainment but it may also affect the overall employment rate). It’s a very nuanced picture and it’s probably disingenuous of some tech leaders to say “we’ll all benefit) but some do believe that will be the future.

It can even help you learn to tie a noose.

https://duckduckgo.com/?q=chatbot+suicide

14-year-old Sewell Setzer III https://apnews.com/article/chatbot-ai-lawsuit-suicide-teen-a...

16-year-old Adam Raine https://www.npr.org/sections/shots-health-news/2025/09/19/nx...

The winner takes all thesis would be that like TSMC, the capex of competing in this field keeps growing until only one vendor can both raise sufficient capital to compete and effectively execute with that capital. OpenAI doesn't need to be the first to stop raising money and go profitable, they need to be the last vendor to go profits first.

It's not impossible that the crash will hit all companies at the same time and they might all stop training.

Yeah they are all over the place and this should be a huge red flag.

Some marginal investors know this but they are okay because the music is still playing - but when they think its time to leave the bubble will pop.

People seem to forget that its not about whether or not its actually a bubble, its really about when will certain people who set these stock prices for valuations, decide its time to exit and take their profit.

So it is regulatory arbitrage. I think many of the critics never contested that crypto is good for doing crime. The critics just also think that either those crimes should continue to be prevented via the financial system or the financial system should be deregulated for all without a crypto backdoor.

A Money Market Fund gives you interest if you are able to access it.

This is kind of a pattern:

1. There is some regulation that is inefficient ( e.g. taxi medallions, KYC, copyright protection ...)

2. New technology comes about which allows startups to claim that they have invented a new area that should be regulated differently

3. Turns out (2) is not true and new technology can easily be mapped to existing regulation but it would look bad for the regulator to take away the punchbowl

4. There is some down-turn (bubble pops) and the regulator takes away the punchbowl OR investors have accumulated so much money/power that they corrupt the government to have new rules for their businesses

In some Hollywood reality, probably.

(Like those people that casually assume that everyone has a therapist or a lawyer.)

> AI is more useful than social media.

Communication is a fundamental human need.

Generating slop isn't.

If you're ever been to a third world country then you'd see how this is completely untrue. The dotcom boom has revolutionized the way of life for people in countries like India.

Even for the average person in America, the ability to do so many activities online that would have taken hours otherwise (eg. shopping, research, DMV/government activities, etc). The fact that we see negative consequences of this like social network polarization or brainrot doesn't negate the positives that have been brought about.

Don’t say that, you’ll hasten the Antichrist!

> The Dotcom boom was probably good for everyone in some way, but it was much, much better for the extremely wealthy people that have gained control of everything.

You are describing platform capture. Be it Google Search, YouTube, TikTok, Meta, X, App Store, Play Store, Amazon, Uber - they have all made themselves intermediaries between public and services, extracting a huge fee. I see it like rent going up in a region until it reaches maximum bearable level, making it almost not worth it to live and work there. They extract value both directions, up and down, like ISPs without net-neutrality.

But AI has a different dynamic, it is not easy to centrally control ranking, filtering and UI with AI agents. You can download a LLM, can't download a Google or Meta. Now it is AI agents that got the "ear" of the user base.

It's not like before it was good - we had a generation of people writing slop to grab attention on web and social networks, from the lowest porn site to CNN. We all got prompted by the Algorithm. Now that Algorithms is replaced by many AI agents that serve users more directly than before.

- cloud gaming service? :D

this is a good point, and it would be interesting to see the relative value of this building and housing 'plumbing' overhead Vs the chips themselves.

I guess another example of the same thing is power generation capacity, although this comes online so much more slowly I'm not sure the dynamics would work in the same way.

This is true. It’s probably 2-3 times as long as a GPU chip. But it’s still probably half or a quarter of the depreciation timeline of a carrier fiber line.

In 1998 to transfer a megabyte over telephone lines was expensive and 5 years later is was almost free.

I have not seen the prices of GPUs, CPU or RAM going down, on the contrary, each day it gets more expensive.

I vibe coded 5 deep ML libraries this month. I'm an MLE by trade and it would have taken me ages without AI. This wasn't possible even a year ago. I have no idea how anyone thinks the models haven't improved

Not even remotely. In LLM land, the progress seems slow the past few years, but a lot has happened under the hood.

Elsewhere in AI however progress has been enormous, and many projects are only now reaching the point where they are starting to have valuable outputs. Take video gen for instance - it simply did not exist outside of research labs a few years ago, and now it’s getting to the point where it’s actually useful - and that’s just a very visible example, never mind the models being applied to everything from plasma physics to kidney disease.

If those statistical models are helping you do better research, or basically doing most of it better than you can, does it matter? People act like models are implicitly bad because they are statistical, which makes no sense at all.

If the model is doing meaningful research that moves along the state of the ecosystem, then we are in the outer loop of self improvement. And yes it will progress because thats the nature of it doing meaningful work.

Uber didn't invent anything, they drove taxis out of business then jacked up the prices, squeezed the workers, and now everybody is riding around in regular people's crappy cars for more than a cab cost. And now I need a phone and to maintain my relationship with these two crappy companies rather than to wave at the street (which is what I used to do to get a ride.)

That was literally what everybody said would happen.

Uber sold something like $50 billion in equity and debt before it went public, and although they're profitable now, to me it doesn't seem like they have answer to Waymo coming up fast in the rear-view mirror. I think Uber is still a scam, just one where the earlier investors fleeced the later ones who are never going to see the returns they paid for.

Faculty at the college I work for are kicking and screaming about AI and students using AI. They don't want to use new tools so they're just trying to outright ban students from using them.

One smug English faculty said, "well it's not that hard. You just look for dashes in their writing."

I responded with, "you know you can just tell it not to use those, right?"

Blank stares.

I hate it.

Based on what evidence? The last few times this happened the whole world benefitted.

Doubtful. More likely that there will be robocops to stop you from loitering in the scenic places.

No they don't want that, because it could lead to an uprising against them. They want us provided with the essentials in exchange for being dutiful workers, so we have something to lose.

Everyone should live in pods stacked together, eat insects, not drive our own automobiles around or fly places, we should be able to get our entertainment and everything to keep ourselves happy from their subscription entertainment services. Basically we are to consume as little as possible to barely keep ourselves alive and sane while they sail themselves around to pat one another on the backs at their climate and economic conferences on their billion dollar luxury yachts.

Actually no that would be stupid they don't have the time or patience to sail their yachts around. They have crew for that. They will fly in one of their handful of private jets and have the yacht meet them there.

Also AOL was a mix of dialup provider and Dotcom service. There were many other popular examples of such.

>dotcom bubble had tons of activity, too.

The difference now is that this is all (or mostly) idle cash being invested. The massive warchests built up by FAANG over the last decade are finally being deployed meaningfully rather than sitting in bonds or buying back stock. Much different scenario than companies with non-viable business models going IPO on a wish and a dream.

They evolved into Instagram(8 billion), WhatsApp(20 billion) and FB(trillion)

Yeah, Sam Altman’s recent one saying that his LLMs would “cure cancer” if they built enough data centres was also a winner.

>You can download a LLM, can't download a Google or Meta.

You can download a model. That doesn't necessarily mean you can download the best model and all the ancillary systems attached to it by whatever service. Just like you can download a web index but you probably cannot download google's index and certainly can't download their system of crawlers for keeping it up to date.

(For those who missed it: https://www.theverge.com/ai-artificial-intelligence/785407/p... )

But why not measure $ per intelligence. In 2020 you'd need a billion dollars to get your computer to write good code, now it is practically free.

In 1998, 16 MiB of RAM was ~$200, in 2025, 16 GiB of ram is about $50. A Pentium II in 1998 at 459 MHz was $600. Today, a AMD Ryzen 7 9800X can be had for $500. That Ryzen is maybe 100 times as powerful as the Pentium II. What's available at what price point has changed, but it's ridiculous how much computing I can get for $150 at Best Buy, and it's also ridiculous how little I can do with that much computing power. Wirth’s law still holds: software is getting slower more rapidly than hardware is getting faster.

> If the model is doing meaningful research that moves along the state of the ecosystem, then we are in the outer loop of self improvement.

That's a lot of vague language. I don't really see any way to respond. I suppose I can say this much: the usefulness of a tool is not proof of the correctness of the predictions we make about it.

> And yes it will progress because thats the nature of it doing meaningful work.

This is a non sequitur. It makes no sense.

And I never said there's anything bad about or wrong with statistical models.

> Uber didn't invent anything

They didn't. They innovated, practically implemented ideas that resulted in the introduction of new goods and services [1]. This is a meaningful difference.

Uber didn't invent anything. But they did pull ridesharing out of a hat.

[1] https://en.wikipedia.org/wiki/Innovation

It completely changed an industry. Just because you don't like the current version doesn't mean it wasn't a major innovation. Uber was just the main player in scaling and marketing a new model across the entire globe, which was a huge and costly endeavor.

LLMs have already changed how software can be written and a thousands of other business/consumer usecases, these companies are just battling it out and finding the most profitable niches. It will be a major business for a long time and the technology will mature and plateau pretty quickly. If R&D doesn't scale economically, it will just slow down and existing models will be heavily optimized to be cheaper to run.

The dot com boom resulted in very few real industries and comparing the two is not very useful.

I think you're definitely in the minority. For most of the world, Uber/Grab/Bolt have made transportation cheaper, safer, more convenient, and more comfortable.

> rather than to wave at the street (which is what I used to do to get a ride.)

Ummm taxis aren't everywhere like NYC or something. Broadly speaking Uber will pick you up in an arbitrary place and take you anywhere.

Uber definitely screwed the workers and probably existing taxi companies, but for the users it was a huge W, at least in many parts of the world. Taxi companies are notoriously scammy and it seems to be a very universal experience.

Uber is still >$10bn net negative over its lifetime. Claude estimates about –$19bn.

It's still quite possible that Uber doesn't make that up in the near future.

If you didn't write it, you don't deserve credit for it. You don't learn by outsourcing your work to a piece of software.

When have we ever had a push to full automation?

At that stage, owners will very likely have enough drones and robots at their command to not need to worry about petty things like flesh and blood uprisings.

That would be https://marshallbrain.com/manna

Every strata of society has been getting richer and richer along with technological progress. What makes you think that would change now?

I can’t speak for others countries but here in the UK special “VIP lanes” were used to steal billions using fraudulent PPE contracts.

I’m not just talking about Baroness Mone either. PPE Medpro was the tip of the iceberg.

Ib the US, 1995 by far!!!

1995 (and I was 22 in 1995)

Expand the thinking to include impact on developing countries, the poor, minority groups who have few people like themselves in their local area, etc.

I’ll grant that for comparatively wealthy, privileged people who were always going to have an easy time (which frankly include me), the internet has been a mixed bag.

But for the kids growing up in comparatively poor countries, who can now access all of the world’s information, entertainment, and economy.. I think it’s a pretty clear win.

I expect AI will be similar: perhaps not a huge boon to the best off, but a substantial improvement for most people in the world. Even if we can sit back and say “oh, but they also get misinformation and lower quality YouTube content”

> This is incomplete in key ways: it only increases knowledge if people practice information literacy and validate AI claims, which we know is an unevenly-distributed skill. Similarly, by making it easier to create disinformation and pollute public sources of information, it can make people less knowledgeable at the same time they believe they are more informed. Neither of those problems are new, of course, but they’re moving from artisanal to industrial scale.

Totally agree with this

>The more successfully businesses are able to remove or deprofessionalize jobs, the smaller the pool will be of people who can afford to build skills, compete with those businesses, or contribute to open source software

I'm mixed on this, ultimately its the responsibility of individuals to adapt. AI makes people way more capable than they have ever been. It's on them to make something of it

> but it’s worse for society since it accelerates the process of centralizing money and power.

I'm not sure this is true, it enables individuals like they never have been before. Yes there are the model infrastructure providers, but they are in a race to the bottom

Probably. The the internet was more clear cut. Moores law all the things. Governments investing in pipe etc.

AI harder to tell. Will 2026 models kick 2025 ass or just be slightly better. Who knows.

And we aren't also in some other bubble. From rest of the world those compensation levels sound insane.

Earning $60k/yr would put you in the global 1%.

Pretty sure discord and slack are not useless…

It was founded 1998, very much dotcom era, and had no significant web 2.0 products for years after that.

I think you’re putting too much weight on cost (time, money), and not enough weight on “quality of life”, in your analysis.

For sure, we can shop faster, and (attempt) research and admin faster. But…

Shopping: used to be fun. You’d go with friends or family, discuss the goods together, gossip, bump into people you knew, stop for a sandwich, maybe mix shopping and a cinema or dinner trip. All the while, you’d be aware of other peoples’ personal space, see their family dynamics. Queuing for event tickets brought you shoulder to shoulder with the crowd before the event began… Today, we do all this at home; strangers (and communities) are separated from us by glass, cables and satalites, rather than by air and shouting distance. I argue that this time saving is reducing our ability to socialise.

Research: this is definitely accelerated, and probably mostly for the better. But… some kinds of research were mingled with the “shopping” socialisation described above.

Admin: the happy path is now faster and functioning bureaucracy is smoother in the digital realm. But, it’s the edge cases which are now more painful. Elderly people struggle with digital tech and prefer face to face. Everyone is more open to more subtle and challenging threats (identity theft, fraud); we all have to learn complex and layered mitigation strategies. Also: digital systems are very fragile: they leak private data, they’re open to wider attack surfaces, they need more training and are harder to intuit without that training; they’re ripe for capture by monopolists (Google, Palantir).

The time and cost savings of all these are not felt by the users, or even the admins of these systems. The savings are felt only by the owners of the systems.

Technologgy has saved billions of person-hours individual costs, in travel, in physical work. Yet, wemre working longer, using fewer ranges of motions, are less fit, less able to tolerste others’ differences and the wealth gap is widening.

With telecom, we benefited from skipping generations. I got into a telecom management program because in 2001-ish, I was passed by on a village street by a farmer bicycling while talking on his cellphone. Mind you my family could not afford cellphone call rates at the time.

In fact, the technology was introduced out here assuming corporate / elite users. The market reality became such that telcos were forced kicking and screaming to open up networks to everybody. The Telecom Regulatory Authority of India (back then) mandated rural <> urban parity of sorts. This eventually forced telcos to share infrastructure costs (share towers etc.) The total call and data volumes are eye-watering, but low-yield (low ARPU). I could go on and on but it's just batshit crazy.

Now UPI has layered on top of that---once again, benefiting from Reserve Bank of India's mandate for zero-fee transactions, and participating via a formal data interchange protocol and format.

Speaking from India, having lived here all my life, and occasionally travelled abroad (USAmerica, S.E. Asia).

We, as a society and democracy, are also feeling the harsh, harsh hand of "Code is Law", and increasingly centralised control of communication utilities (which the telecoms are). The left hand of darkness comes with a lot of darkness, sadly.

Which brings me to the moniker of "third world".

This place is insane, my friend --- first, second, third, and fourth worlds all smashing into each others' faces all the time. In so many ways, we are more first world here than many western countries. I first visited USAmerica in 2015, and I could almost smell an empire in decline. Walking across twitter headquarters in downtown SF of all the places, avoiding needles and syringes strewn on the sidewalk, and avoiding the completely smashed guy just barely standing there, right there in the middle of it all.

That was insane.

It seems the crux is that we needed X people to produce goods, and we had Y demand.

Now we need X*0.75 people to do meet Y demand.

However, those savings are partially piped to consumers, and partially piped to owners.

There is only so much marginal propensity to spend that rich people have, so that additional wealth is not resulting in an increase in demand, at least commensurate enough to absorb the 25% who are unemployed or underemployed.

Ideally that money would be getting ploughed back into making new firms, or creating new work, but the work being created requires people with PHDs, and a few specific skills, which means that entire fields of people are not in the work force.

However all that money has to go somewhere, and so asset classes are rising in value, because there is no where else for it to go.

The data centers built in 1998 don't have nearly enough power or cooling capacity to run today's infrastructure. I'd be surprised if very many of them are even still in use. Cheaper to build new than upgrade.

Eh not really. Maybe retro cloud gaming services. But games haven't stopped getting more demanding every year. Not only are the AI GPUs focused on achieving clusters with great compute performance per watt and dollar rather than making singular GPUs with great raster performance; even the GPUs which are powerful enough for current games won't be powerful enough for games in 5 years.

Not to mean that we're still nowhere near close to solving the broadband coverage problem, especially in less developed countries like the US and most of the third world. If anything, it seems like we're moving towards satellite internet and cellular for areas outside of the urban centers, and those are terrible for latency-sensitive applications like game streaming.

Even if the building itself is condemnable, what it took to build it out is still valuable.

To give a different example, right now, some of the most prized sites for renewable energy are former coal plant sites, because they already have big fat transmission lines ready to go. Yesterday's industrial parks are now today's gentrifying urban districts, and so on.

> This wasn't possible even a year ago.

My experience has been that it was. I was using AI last year to build ML models about as well as I have been this year.

I'm not saying AI isn't useful, just that the progress certainly looks to be sigmoid not exponential in growth. By far the biggest year for improvement was 2022-2023. Early 2022 I didn't think any of the code assistants were useful, by 2023 I was able to use them more reliably. 2024 was another big improvement, but I honestly haven't felt the change (at least not for the better).

Some of the tooling may be better, but that has little do to with exponential progress in AI itself.

> the progress seems slow the past few years, but a lot has happened under the hood.

The claim is "exponential" progress, exponential progress never seems "slow" after it has started to become visible.

I've worked in the research part of this space, there's neat stuff happening, but we are very clearly in the diminishing returns phase of development.

I think Uber invented (or at least made widely available) taxi-by-app. Having scores of cab companies, accessed by telephone call, was supremely user-hostile.

If the cab companies had gotten together on an app, they might have shortcut Uber and all of its many dubious practices. They finally are starting to but it's much too late.

Uber is way cheaper and better than any cab ever was.

It's sad. I use dashes in my writing all the time. I've had to consciously pause and make an effort not to use them in my emails now.

The writing is a mandatory part of the learning process. Outsourcing it to tools is skipping and cheating. Nothing new here. It was the same when the tool was your buddy or upwork before LLMs.

Banning advanced graph calculators on undergrad math exams is not because "they don't want to use new tools", either.

Based on extensive academic research on trickle-down economics, in particular looking into the evolution of real wages of different sectors of population, since 1980s.

See the work of recent Nobel prize laureates in economics. Many argue for redistribution and investment back to the society.

For sure, but most bridges aren't in scenic places and robocops will still be limited in numbers, so no worries, just ask your LLM assistance in case of doubt under which bridge you won't get evicted from. Altough, that would probably be illegal advice, but they can try to help you with therapy should you end up in such a situation. Not the top model of course, but hey, it is a benefit.

don't forget the dismantling of the local community and family identity to make people reliant on the government preventing resistance by division.

I'd actually love to eat insects but to my dismay I found out they're actually really expensive...

How will they steal added value and get richer if masses consume as little as possible?

> Everyone should live in pods stacked together, eat insects, not drive our own automobiles around or fly places, we should be able to get our entertainment and everything to keep ourselves happy from their subscription entertainment services.

Yes of course capitalists love when economy is bad. Sorry, these dystopic visions do not pass even simplest smell test.

For just $19.99/mo no doubt.

Energy, raw materials, and logistics still remain. I don't think we'll ever get to a place where there isn't some input to a production process that is not infinite and free.

Cite?

How so?

Given my skillset at age 22? Yeah, I'll take 1995. I was old enough to grow up hearing how great the world was going to be if I learned computer programming just to enter the job force at the start of the dotcom bubble burst. 1995 would have been a major upgrade.

Also, knocking that almost decade off my birthday would assure that I spent most of my adult life with the luxury of thinking that energy didn't have negative externalities that were being forced on later generations.

We had Chomsky-esq "any major world power is kind of fascist if you think about it" instead of literal talk by politicians about putting people in camps if they don't like your diet or country of origin.

TV was pretty bad I guess but music was great and I read more back then.

There was a lot of huffing and puffing about gang violence. I grew up on the street the local gang named themselves after and it only marginally touched my life at all.

Housing was dirt cheap, food was dirt cheap, gas was dirt cheap. There was undeveloped land everywhere around the city I live in and it gave a general sense of potential.

What exactly was so bad about the 90's?

Are you being objective or just romanticizing the past?

Just to use your example: YouTube is filled with talented writers and storytellers, who would have never been able to share their content in the past. *And* the traditional media complex is richer than ever.

I don’t think average quality matters. Just what you want to consume.

If anything, I’d be more open to the opposite argument. Media is so much richer and more engaging that it actually makes our lives worse. The quality of the drugs is too high!

The political instability that came with social media isn't great, either.

Undeniably better off in every single way. Minimum is that the price of long distance phone calls is now zero, let alone video calls. Being able to speak to family and see them nonstop is incredible.

Does it have to be for the analogy to work?

I need to move back to the Bay

And what is it that you believe about these old money types' knowledge of AI and why is it that you believe it?

> Agreed. So I don't think it's a bubble.

I'm not arguing it's a bubble. I'm arguing it's not going to be a "gigantic benefit" for society.

> Will also be good for consumers in the long-term: much faster pace of drug discovery and new tech generally.

Medicine and tech that knowledge workers won't be able to pay for without a job. I also don't think "new tech" is necessarily a good thing societally.

The calculus is easy, really: AI makes 100% of my income worth less but only decreases the cost of a fraction of my expenditures. AI is bad for anyone that works for a living. That's pretty much all of society except for the top x%.

Recall that the arXiv was established in 1991, many years before the dotcom bubble. Many scientists still use arXiv prominently for research.

That’s true. The permitting is already dealt with, and that’s substantial lead time on any data center build.

Evnn moreso for carrier lines of course. Nimbyism is a strong block on right-of-way needs (except the undersea ones obviously).

> But games haven't stopped getting more demanding every year.

This is not particularly true.

Even top of the line AAA games make sure they can be played on the current generation consoles which have been around for the last N years. Right now N=5.

Sure you’ll get much better graphics with a high end PC, but those looking for cloud gaming would likely be satisfied with PS5 level graphics which can be pretty good.

We continue to see steady progress, look at any benchmark and its basically a linear increase.

Wow really? The agentic coding work that has come out in the last year are super impressive to me.

And before it didn’t seem to understand the fundamentals of Torch well, not well enough to do novel work. Now with Codex in high it absolutely does, and MLE bench reflects that

Indeed, the very dashes the English teacher taught me to use.

Is that still true?

I suspect that's beyond what the Hackernews crowd can cope with contemplating. Billionaires bad except on that subject in which case billionaires good and working class bad.

Soylent green will be affordable when the baby boomers start passing.

Just go outside for a while with a survival book if you are into it.

All the value will be diverted. I don’t know if you noticed, but in the last few years, you’d have come ahead if you gambled on the stock, crypto or commodities market than if you busted your ass.

None of this was ever intended to be cheap for you.

It’s less complicated than that. Externalities are when an individual profits but others pay the price. Think of climate change. And when a small number are so rich they control the government, so there’s nothing to stop them. It’s narrow incentives driving the whole thing.

The economy won't be bad. The numbers will look good for them. Also they aren't capitalists. Try to keep up.

LLMs are a bubble on the back of an AI bubble in a market finally surmounting an enormous single-shot energy surfeit

Gmail and Google Maps were prototypical 2.0.

Discord and slack are evolutions of msn messenger and icq.

Not web chat rooms.

> I think you’re putting too much weight on cost (time, money), and not enough weight on “quality of life”, in your analysis.

"Quality of life" is a hugely privileged topic to be zooming in on. For the vast majority of people both inside and outside the US, Time and Money are by far the most important factors in their lives.

And the term 2.0 was coined in 1999.

Yes, this is true. There used to be a lot of local book stores, for example. Amazon optimized that away, while ruining social fabric.

> Now we need X0.75 people to do meet Y demand.*

This is how GDP/person has increased 30x the last 250 years.

What always happens is that the no longer needed X*0.25 people find new useful things to do and we end up 33% richer.

That kind of extreme poverty juxtaposed to extreme wealth, and all of the social ills that come along with it, have always been a fixture of the American experience. I don’t think it’s a good barometer or whether the USA is in decline when there has long been pockets of urban decay, massive inequality, drug use etc. Jump back to any point in American history and you’ll find something similar if not much, much worse. Even in SF of all places, back in the wild west era gold rush or in the 1970s… America has always held that contradiction.

> partially piped to consumers, and partially piped to owners.

Or, the returns on capital exceed the rate of economic growth (r > g), if you like Piketty's Capital in the Twenty First Century.

One of the central points is about how productivity and growth gains increasingly accrue to capital rather than labor, leading to capital accumulation and asset inflation.

How come? I'd expect that efficiency gains would lower power and thus cooling demands - are we packing more servers into the same space now or losing those gains elsewhere?

From my experience, this is not the case in many european cities. In those places the main benefit seems to be not having to interact with the cab driver.

Taxi-by-app existed pre-Uber, the innovation was making the taxi actual show up. Austin had the apps, and I would order one, and the taxis would get distracted on the way to my house and pick up other fares, so I couldn't go where I wanted. They had every chance to not be outcompeted by Uber, but they couldn't stop being taxis. And, here we are.

The sad difference that makes a difference is that banning graphing calculators from proctored exams is enforceable in practice.

Appealing to honor is a partial solution at best. Cheating is a problem at West Point, let alone the majority of places with a less disciplined culture. It's sad, but true. The fact that you and I would never cheat on exams simply does not generalize.

https://www.armytimes.com/news/your-army/2021/04/18/51-west-...

edit: good on West Point for actually following up on the cheating. I've witnessed another institution sweeping it under the rug even when properly documented and passed up two or three levels of reporting. As an academic director and thus manager of professors this was infuriating and demoralizing for all concerned.

Sure but there's also a continuum of practice between banning outright and letting the students use llm's to just write for them.

Use it as a peer review. Use it during brainstorming. Use it to clarify ambiguous thoughts. Use it to challenge your arguments and poke holes in your assumptions.

>robocops will still be limited in numbers

The population of developed countries is already at least 75% robocops

But the past few revolutions benefitted everyone and we are better off. Look at industrial revolution, digital revolution. Why do you think it is different this time? If trickle down economics don't work, why is world poverty at all time low and consumption at all time high?

People do this to themselves and willingly, no spooky evil capitalists behind the curtains are necessary.

People love money and what they can bring on the table, people often hate each other ie within families stiffed by peer pressure and expectations of mentalities formed in another very different era, people love discovering new countries and cultures. And so on and on.

Ie me - I love my parents, my childhood was normal, only later to find and compare with others to see how such childhood was... abnormally uncommon. But I very much prefer seeing them few times a year only, even though we love when they help with kids. Some of their opinions are very outdated, their ramblings are often out of touch with reality, they tend sometimes to spoil kids (even after setting boundaries), and overall generational gap is absolutely massive. It is a form of freedom. Make that 10x more in much more strict societies where pressure and expectations from parents on kids are massive and then they wonder why kids stay the heck away from them once adults.

And fuck local communities, for every good-hearted neighbor who just wants to socialize and help out and otherwise stays away from one's life, there is easily 5 or 10 who are the epitome of nimbyism, voyeurism or similar hobbies of people with empty lives, clueless on how world and people actually work but always with very strong opinions on everything and will to push those on everybody else.

They love it when the richest people do well. They dont care about how anyone else lives. The poorer other people are better they feel about winning.

Those you call "capitalists" love monopolies as long as they are theirs. They love captured market. They dont care about competition unless it is someone not them competing to provide for them on lower price.

As of now, billionaires dont want or need strong economy as a "middle class and lower class doing good". They want the "our wealth goes up, we are getting tax breaks, if lower class pays for it cool" kind of economy.

It's really not though. There are a few high quality individual creators, but not many.

I wish this argument would die. We're asking for a better future among futures, not a better future compared to the past.

It's like buying a car, receiving a bike, and then being told, "A bike is great because you don't have to walk anymore." If you feel like that's unfair and the response misses the point, that's how that lands. I don't know who, when hearing that, feels better. It feels out of touch and dismissive.

No.

Chats did not evolve to social media.

Facebook is a descendant of hotornot not icq or chats.

Ws was seen as free sms.

And create a theory of quantum gravity.

Setting aside time, is money not downstream from quality of life? Meaning, in a better world one might not need to care as much about money? I believe that time and quality of life are congruent - good quality of life means control over one’s own time.

Costs (time/money) are metrics to help analyse your situation/progress. They are not quality of life.

If you had a huge pile of money but still lived in a shack in a slum, you’d still have a terrible quality of life.

If you argument were true, and people are saving time (or money) due to these new systems, why is the wealth gap widening?

Right, but the term and the concept didn't really take off until years later. From https://en.wikipedia.org/wiki/Web_2.0:

> The term "Web 2.0" was coined by Darcy DiNucci, an information architecture consultant, in her January 1999 article "Fragmented Future" [...] her "2.0" designation refers to the next version of the Web that does not directly relate to the term's current use.

> The term Web 2.0 did not resurface until 2002.

Google's first big Web 2.0 products were GMail (beta launched in 2004, just before Google's IPO) and Google Maps (2005).

RIP literacy in America. We hardly knew ye.

Yeah, I sort of recounted a stark memory. That juxtaposition was a bit too much.

However, it wasn't just that, and the feeling has only solidified in three further visits. It isn't rational, very much a nose thing, coming from an ordinary software programmer (definitely not an economist, sociologist, think tank).

"we end up."

It's actually, "they end up" and the 33% gains you're talking about aren't realized en masse until all the coal miners have black lung. It's really quite the, "dealy" as Homer Simpson would say. See, "Charles Dickens" or, "William Blake" for more. #grease

Power limitations are a big deal. I haven't shopped for datacenter cages since web 2.0, but even back then it was a significant issue. Lots of places couldn't give you more than a few kw per rack. State of the art servers can be 2kw each, so you start pushing 60kw per rack. Re rigging a decades old data center for that isn't trivial. Remember you need not just the raw power but cooling, backup generator capacity, enough battery to cover the transition, etc.

It's hugely expensive, which is why the big cloud infrastructure companies have spent so much on optimizing every detail they can.

It's just an educated guess, but I expect that power density has gone up quite a bit as a form of optimization. Efficiency gains permit both lower power (mobile) and higher compute (server) parts. How tightly you pack those server parts in is an entirely different matter. How many H100s can you fit on average per 1U of space?

Yes - blades-of-servers replacing what was 2 or 3 rack mount servers. Both air exchange and power requirements are radically different in order to fill that rack as it was before.

Not only did taxi by app exist way earlier, it was also wide spread ... in some big countries.

I dont see anything reason at all to expect West Point to have less cheating then anywhere else. Army people dont commit crimes less then civilians either.

Time to crack out the faraday cage!

the last bit is only true under individualised consumerism. its when everybody is like a spoiled child is where solidarity goes to die.

The poorer people are that are still able to work, the cheaper labor is. It's really that simple.

It's a question of how far the contagion goes - the Dotcom bubble trashed the NASDAQ and rumbled the world a bit, but the "big stocks" weren't directly involved or affected much; the subprime mortgage bubble shook the very foundations of the banking world.

Will the contagion be limited to a few companies, or will the S&P crumble under the bubble's collapse?

I don't think it is fair to characterize AOL as a bubble baby- they were a thing before that

i have to say I'm a little disgusted by these statements. LLMs are useful for many problems, but is there really a conceivable path of them making progress into fighting the countless cancers tormenting humanity?

LLMs are the air in the AI bubble.

Machine learning is a perfectly valid and useful field, traditional ML is super useful and can produce very powerful tools. LLMs are fancy word predictors that have no concept of truth.

But LLMs look and feel like they're almost "real" AI, because they talk in words instead of probabilities, and so people who can't discern the distance between and LLM and AGI assume that AGI is right around the corner.

If you believe AGI is right around the corner and skip over the bit where any mass application of AGI to replace workers for cheaper is just slavery but with extra steps, then of course it makes sense to pour money into any AI business.

Discord is literally a web chat room. It operates over the web. Its interface is a re-branded web browser. You chat in it. There are rooms (although I suppose people call them "channels," implying a conceptual link to IRC).

Gmail's beta started in 2004, and Google Maps launched in 2005. Google's IPO was in 2004, so Web 2.0 products played no part in its enormous early growth.

that makes absolutely no sense

Independent book stores are booming in quantity since the initial Internet crash.

Be careful about making narratives that don’t line up with industry data.

There’s a lot of brick and mortar retail going on. It just doesn’t look like the overbuilt mall infrastructure of the 1970s-1980s.

Yep, that’s the source of the point. The effort is in finding a way to make it easy to convey. Communication of an idea is almost as critical as its verification now.

You don't talk to your drivers? I've had a lot of genuinely good conversations in my rides. I always at least feel it out and see how much the driver wants to talk but it feels wrong to not say anything

> the innovation was making the taxi actual show up

Uber regularly doesn’t show up, just playing with “4… 3… 6… minutes left”. I always have to wait half an hour at a certain location, with a message “Just book a few minutes before your trip.”

I’ve had a friend miss his flight because of Uber not showing up!

Are the examples you mentioned actually banned, as opposed to not actively used in classes? I'd think even where LLMs don't belong in the curriculum, self-studies in various forms as complimentary to the curriculum wouldn't be any of the teachers business?

BTW if you haven't, I encourage reading this.

https://zenodo.org/records/17065099

> Look at industrial revolution

I really don't see how one can separate the industrial revolution from colonialism, considering we have chiefs of government in colonial countries on the record saying that colonies are a necessary outlet for industrial goods [1].

Once you've established that link, it's hard to explain that "everyone" benefitted from the industrial revolution.

Even disregarding that, the working conditions created by industrialization allowed for situations that can hardly be described as "beneficial" [2][3][4].

> digital revolution.

[5] provided without comment.

---

[1]: for instance, Jules Ferry: https://www2.assemblee-nationale.fr/decouvrir-l-assemblee/hi...

[2]: https://en.wikipedia.org/wiki/Bisbee_Deportation

[3]: https://en.wikipedia.org/wiki/Ludlow_Massacre

[4]: https://en.wikipedia.org/wiki/Coal_Wars

[5]: https://en.wikipedia.org/wiki/PRISM

Erm, what qualifies as a robocop to you?

But kids in those strict societies stay with parents more, help them and treat them better.

Meanwhile, you dont care about them except for them being potential resource of free work.

And I mean, you idea of local comunity is all about other people doing free work for you and then tolerating your peculiarities with no reciprocation.

So, I dont think that is much of an argument here.

As noted, this refers to Reagan policies and onward.

Nor I dont think Coget, GX or Level-3 were dotcom companies.

in the meantime USA and all the world is going actually all-in AI..

Without those sorts of revolutionary claims, would they still be receiving enough funding to keep the lights on?

In the 2000 every single webchat room was connected to irc.

Discord is not.

And energy is the air in the economy at large. The Eliza effect is not the only bias disposing people to believe AGI is right around the corner. There are deeper assumptions many cling to.

I agree about ML.

msn messenger became Skype wich then became ms teams.

Gmail had a fundamental part in Googles IPO.

Two decades ago, in the Bay Area we used to have a lot of books stores, specialized, chains, children's, grade school, college slugbooks, etc. Places like Fry's had a coffee and a book store inside. The population grew, number of book stores went down to near zero.

Im not sure why you got downvoted for that.

Ive noticed similar shenanigans with Uber eats - they are definitely playing around with stuff to increase profitability its clear as day.

I'm talking about the benefit to whoever is using them. Once you learn it's more expensive, there's little logic to keep on using it unless for the reason stated.

Correct. The faculty at my institution do not want any use of LLM or 'AI' technology. End of sentence. If they learn that a student used an LLM, regardless of how it was used, they send a formal academic discipline complaint to administration. It's a fucking joke.

I'm not saying throw the doors open and let loose. I'm saying that we need to find places where using these tools makes sense, follows a sense of professional ethics, and encourages (rather than replaces) critical thinking.

And the problem with your cited paper is that people who kick and scream the loudest about this at my institution (again, this is just at mine and is in no way indicative of any other institution) are the ones who have not updated their courses since I was in college. I mean that quite literally. I attended the institution I currently work at. Decades later and I could turn in papers that I wrote for their classes my freshman year and pass their classes.

Three of them sent me that same linked article. But instead of seeing the message "we need to think about how to use these things responsibly" they just read "you can do what you've done for years and nothing needs to change."

That "research" article isn't as impactful as the faculty at my institution thought.

I'm all for the thoughtful integration or rejection of these technologies based on sound pedagogical practices rooted in student learning theory. At my institution, and I want to stress n=1, they literally do not want to take time updating lessons, regardless of the reason. Llm's are just a convenient scapegoat right now.

I would argue that it's more unethical to not update your classroom lessons in over 2 decades than it is to use llm's to supplement learning.

Being robotic and policing.

Maybe I phrased wrongly (not a native speaker) but what you say is incorrect. I am not free coasting on parents, in contrary my first years all my earning went into providing them a good home for rest of their lives, while living in tiny rental rooms. You have no idea from sort of poor background I come from, most western kids have no clue. It was a massive improvement of QoL for them that they would never be able to afford themselves and they still appreciate it massively. Since we live 1500km from them they help with kids literally few hours per year, not a burden for anybody involved.

But freedom to chose with whom we spend our time is a thing, no matter how much people like you try to force their righteous values that are the only proper true way (TM) on everybody else. I am old enough and over time met plenty of folks like that, be it religious or other forms, they are at the end the same as your comment.

> And I mean, you idea of local comunity is all about other people doing free work for you and then tolerating your peculiarities with no reciprocation.

When it seems to you like everyone around you is the problem, you may actually be the problem.

My framing device: if you had LLMs in the 1500s, how would that help Copernicus determine the orbits of the planets? Maybe through dumb chance, but creating a well reasoned model of the universe required new observations and the ability to interpret the data from a different point of view.

I just need to read it one of these days. I keep reading parts of it but never sitting down with it.

Places that luckily avoided colonialism haven’t improved a zilch.

What percent of the population in places which experienced the industrial revolution would be better off if they time-travelled back 200 years? 1%? 0.2%?

Even after Reagan's policies how do your points hold?

> I struggle to articulate the way that abundant cheap goods lead to that outcome.

It has nothing to do with how cheap the goods are

The problem is that at some point people won't be able to afford literally anything because all, and I mean literally all, of the wealth will be hyper concentrated in a super small percentage of the population

All the 1500s and earlier data that such an LLM would have to have been trained on would lead to an LLM that wouldn’t ever suggest a heliocentric solar system. That LLM might even say he was heretical or refuse to give an answer to anything that led to it saying that the earth wasn’t the centre of the universe. So no help at all.

Interesting framing. Although I assume all the observations had been done already. It was more about being bold enough to investigate a line of thought that wasn't obvious or popular at the time and proving it convincingly.

They already had many "explanations" and models for why the planets were seemingly moving back and forth in the sky during the year. Their models were more complicated than necessary simply because they didn't want to consider the different premise.

Oh hell, don’t make me feel bad. I’ve got a meager capacity to trudge through major papers. I cheer you on.

You're in luck! A couple years ago he released an "abridged version" of sorts. A Brief History of Inequality is the name. Much more accessible than the 700 pages of Capital in the 21st Century.

https://en.m.wikipedia.org/wiki/Special:BookSources/97806742...

More likely the LLM would condescendingly tell him that he was wrong and that the earth was clearly at the centre of the universe.

> in places which experienced the industrial revolution

People experienced the industrial revolution everywhere.

I suspect, when you think "places which experienced the industrial revolution", you think about a small subset of areas where some development happened as a result of that, likely the areas where industrialists lived.

But you would also have to consider other places' experience of industrialization. For instance, Congo under EIC colonial rule did experience industrialization - it was the place where industrial amounts of rubber were harvested to allow for plants elsewhere to produce joints, pipes, motor belts, etc. It's not really hard to believe that, had Congo not experienced that, its citizen would almost certainly have been better off now.

> Places that luckily avoided colonialism haven’t improved a zilch

Thailand and China would like a word.

Because we are counting people's wealth in significant fractions of a trillion dollars and young people around the world cannot afford homes no more?

I would say that has more to do with the decline in musicals involving small numbers of people doing choreography, and the current movie system de-prioritizing dance as an important skill. The highest grossing musical movie happened in 2024 with Wicked, and the second half of that movie is probably going to do the same thing next year.

Does Congo lack electricity, modern medicine, and air conditioning?

If the industrial revolution has made their lives worse, it's a double-whammy because they are forced to suffer almost twice as long, as their life expectancy at birth has approximately doubled since 1870.

https://www.statista.com/statistics/1071021/life-expectancy-...

Depending on the time of creation, it might be impossible for the LLM NOT to tell him he's wrong and the earth is the center of the universe.

Prices of houses increased but others decreased. Life expectancy increased, consumption increased.

So would Liberia and Ethiopia.

“Phones not homes.”

So we agree that it is more complex than:

colonized => successful

not_colonized => unsuccessful

Consumption of overpriced crap, a.k.a. the lipstick effect [1].

[1] https://en.wikipedia.org/wiki/Lipstick_effect

Daily TV News is limited by travel times. If some story breaks finding the right person and getting them on an airplane and then into a studio can be impractical.

I just think if both are using zoom calls it's a bit unfair to make youtube the bad guy for also doing so.

Okay, but why would we die on the vine? Wouldn’t we just… make a parallel economy without the AGI? The world works today without AGI.

> Okay, but why would we die on the vine? Wouldn’t we just… make a parallel economy without the AGI? The world works today without AGI.

Because you need things they want. Like why would they spare the electricity to heat your home, when it could go to "better" use powering a few dozen GPUs serving a billionaire? Why would they spare the land for you to grow food, when they could use it to build ziggurats dedicated to their power (or whatever else is their whim)?

The market sends the scarce resources to those with the most money.

Or use any of the wonders of military technology invented in the last 200 years to take back society. These fuckers have lived in prosperity for so long that they don't even think it is possible. But so many people throughout history thought they were untouchable until the masses decided they had enough of their shit. Kings with professional armies, full plate knights, men with cannons, mens with guns, machine guns, bombs, planes, tanks, helicopters, and yet at the end of the day when enough random people are pissed off enough the mass of people with nothing to lose are the ones who "win" in the end with the powerful dead or hiding.

Parallel economies are banned. Try making a new currency in the US.

They'll just make some more poor to kill.

Economies can work without currencies. It's a little inconvenient, but bartering/trading goods for services was common in the depression when nobody had any cash.

When unrest happens the military sides with the ones that give better hopes of keeping the stream of money that funds the army flowing.

It sides with the poor only if the powerful (gov) are hopelessly inept at gathering money. If there's a chance that current civilian power can reform and keep collecting the money from the people and funding the army then the army sides with them and help quell the rebellion.

Those who hold kinetic power will never side with poor against the rich.

Did we just ignore the last fifteen years of crypto?

You are simply incorrect.[1] Starship will change it all if it succeeds. The space sector is abuzz with the possibility of orbital refueling and the opportunities it will open up, eg [2]

[1] https://orbitalindex.com/archive/2023-03-15-Issue-210/

[2] https://forum.nasaspaceflight.com/index.php?topic=50157.0

" People will always be able to supply labor at costs below the price of materials for robots."

Why? What critical materials do robots need that will always be more expansive than raising a human?

Also, from the point of "the rich" - the benefit of a robot is, that it will (stupidly) do as command, unlike a human. They don't have a family they want to take care first.

No I had the speculative ponzi front of mind when making that comment.

Governments love crypto because it lets you seize lots of money from criminals across borders. And it is legal gambling where you can tax the winnings without reimbursing the losers (unless they can offset it but most probably can not)

Not always .. see the peasants revolt of 1381, French Revolution in 1789-1793, Bolshevik Revolution in 1917, etc.

Interestingly, new ad-hoc currencies pop up all over in times like that:

https://www.armstrongeconomics.com/history/europes-economic-...

https://collectingpapermoney.spmc.org/wiki/Scrip_Issued_duri...

Im not talking about the military siding with the people, im talking about the military's side not being enough to prevent the people from rising up and taking down whoever is in power. Militaries require logistics, random people do not. Military needs a government or people to follow for direction, masses of angry people do not.

Every time a government or military force has decided they were unstoppable or untouchable, history has proven them wrong. Hell we spent 2 decades in Iraq and Afganistan with the most powerful military and military tech the world has ever known backed by the strongest economy in the world against guys with 60+ year old bolt actions and guns filed out by hand in caves living in mostly desert landscapes, and we still ended up abandoning it because it was too costly. How would the military fair any better against the best armed population in the world with direct access to their supplying economy and logistics networks?

Yeah sure masses of people aren't making aircraft carriers, but you don't need aircarft carriers to win a war at home. We have modern engineering and chemistry text books in every library across the US that will tell you how military technology works and its flaws, what technology you can utilize find or make yourself, and a supply of nearly any material someone could possibly want or need sitting in scrapyards across the nation.

Printing money is lucrative for the printer so any time it might get even a little bit useful and feasible for other parties somebody will start printing.

> im talking about the military's side not being enough to prevent the people from rising up and taking down whoever is in power.

That's sheer fantasy.

https://www.instagram.com/reel/C9AdZVQuITM/?utm_source=ig_we...