So, they’re spending insane amounts of money to develop really snazzy tulip bulbs.
I was curious as to what people thought about this Atlantic article (giftlink) praising Claude and saying AI now leads to astronomical business revenue improvements. Accurate, or hype?
Limited excerpt:
Six months ago, people arguing that AI was a bubble were pointing to real-world facts, Rogé Karma argues, but now “the burden of proof has shifted to the naysayers.”
With Claude Code, a team of autonomous AI agents could take over your computer and, in minutes or hours, complete programming tasks that previously would have taken humans days or weeks. In many cases, the final product required few, if any, human changes. Other companies have since released updates to their own coding tools, such as OpenAI’s Codex and Anysphere’s Cursor, which are considered nearly as impressive as Claude Code. “This really was a step change,” Ethan Mollick, a co-director of the Generative AI Lab at the University of Pennsylvania, told me. “For years now, we’ve been in an era of chatbots that mostly just say things. Now we’ve officially crossed into the era of agents that can actually do things.”
I don’t think the article actually says that — though, to be fair, it was a very wordy article with a lot of buzzwords and seemingly not a whole lot of substance. I might’ve missed that in skimming?
But if not, I think it’s saying that the revenues of AI companies (OpenAI, Anthropic) and infra resellers (like CoreWeave) have increased, not that their tools helped other business users (their customers) dramatically increase their revenue.
Both, no? Like the Dotcom bubble all over again. Certainly there will be fallout and bankruptcies when it pops, but there will also be lasting impacts and success stories.
I think what Google said several years back is still true today: “We have no moat, and neither does [anyone else]”. Anthropic, OpenAI, Google, Meta, the various Chinese companies… they’re all more or less interchangeable. Google at least has substantial custom silicon; everyone else is mostly just buying Nvidia.
Claude Code is great — my entire company uses it many hours a day. But so is Codex. So is Antigravity. And Cursor, and Windsurf, and the dozen copycats coming out every week. And the free open-source tools and models are getting better every month too.
But AFAIK the AI companies are still bleeding money and heavily subsidizing customer usage by burning investor dollars. That’s the hype part. Probably there is some profitable future there, but it would rely on some combination of much higher inference prices, cheaper & less frequent training, fewer competitors, and identifying stable market niches.
For that last part, Anthropic seems to be focusing on the B2B coder market, one of LLMs’ strongest use cases and a reliable income source, but that’s also a much smaller market than the “artificial general intelligence for everyone and everything” that was hyped just a couple years ago.
There is a lot of hype AND some genuine uses worth paying for. It’ll just take some time to figure out which is which and how to charge for it profitably.
At root, the reason is that AI has the potential to be bigger than the industrial revolution.
The industrial revolution allowed biological muscle to be replaced by machine muscle. One of the side effects of that is that economic growth started happening 30+ times faster. Before the industrial revolution, global GDP grew at about 0.1% a year. After it, rates of 3%+ a year were seen. Some places like China grew at 10%+ a year for certain periods, a rate of 100x faster.
Also advances in STEM, medicine, etc occurred far more rapidly after the industrial revolution.
Basically machine cognition has the potential to drastically increase global GDP and speed up the rate of progress in STEM, medicine, etc.
The AI we have now is just the early stage. World GDP is about 125 trillion a year. Advances in AI will cause faster and stronger economic growth in the long run. Endless trillions in new GDP growth per year will arise from advances in AI, which make the hundreds of billions in investments seem like a bargain by comparison.
People may look at all the mistakes AI makes right now, but AI is vastly better than it was 10 years ago, and AI will be vastly better in 10 years vs what it is now.
AI didn’t magically appear overnight, a lot of people have been expecting it to finally arrive and start being useful, we just didn’t know when.
And has no one noticed that the Industrial Revolution also brought us polution and exploitation on scales never before seen, as well as climate change? Not that it doesn’t have its benefits, of course, but I’m just saying maybe we should learn some lessons from the last time we revolutionized human existence and slow the fuck down.
I don’t disagree with you. But history isn’t made by moral people who give a damn. It’s almost always written by the villains.
AI is only going to further concentrate power and wealth — its economies of scale work all too well with the capitalist model — and it won’t matter how many people of conscience philosophically object. We have no power and no money, and ultimately we’re replaceable and expendable, especially if the AI overlords get their way.
Whichever ideology gets the most AI is going to crush every other one, even if it causes mass suffering and extinctions. That’s just how our species has always worked.
Then we jam the gears of the corporate machine with our flesh and bone.
That works for, what, a year or two at a time? Maybe half a decade if you’re lucky? In the meantime, the people who don’t think that way keep building generational wealth and power, and capital keeps concentrating, and technology keeps leapfrogging, and we won’t even make the footnotes.
The world could easily lose a few million (maybe even billion?) people here or there, and I don’t think that will meaningfully slow down “progress” at all — COVID certainly didn’t. We just don’t matter. It’s just a tiny handful of people who control all the actual levers, especially in the sham democracy we’re left with.
Hell, 7 billion people could protest as loudly as they want, but it only takes a handful of billionaires and data centers to completely silent their voices. What voices, even… it’s not like they’d even have a platform to share their thoughts without the billionaire techbros. We live in a world where a single businessperson can hold more wealth and power than entire countries. It’s not just AI and wealth that’s concentrating, it’s all our means of communications, decision-making, gathering, transportation, dating, education, weapons, culture, healthcare… all are falling into the hands of fewer and fewer oligarchs. The world gets a little more dystopian every year.
But hey, in the meantime… ooh, AI cat pics!
It’s being pushed hard first because it’s a scam, and the scammers want to get the maximum amount of money before it collapses. And second because the businesses who buy into the scam desperately want to fire their employees and replace them with “AI”, dreaming of a world where a company consists of nothing but a bunch of machines and a single CEO who hogs the profits all to himself.
While actual AI would change the world, a glorified chatbot isn’t going to do that, and that’s what these so-called “AIs” are.
As for what actually what would happen if we actually got “true” AI? What I expect would likely happen is that the billionaires would have their companies build the AIs, then be completely blindsided when soldiers walked into their office and shot the billionaires in the face as the government took control of everything. Money loses to brute force.
If they actually got their way I expect that Earth would end up like Isaac Asimov’s Solaria. A world with a handful of overlords living in isolation served by legions of robots, and nobody else alive.
I was talking to my elderly mother this morning. She was distressed because her online grocery orders’ substitutions are chosen by AI, and are universally unacceptable, and don’t take her notes into account.
It occurs to me that AI is very much like slavery: workers you don’t have to pay. Unlike actual slavery, AI is like enslaving people who are inherently stupid. I expect business would rather eat the costs of incompetence, especially if the competition is in the same boat, than pay labour. Especially since nothing is measuring all the potential customers who turn away in frustration BECAUSE of AI.
AI is now enabling drone swarms to self-coordinate targeting of enemy forces, without depending on jammable direction from human operators. Never mind economic power, that’s raw military power. No military that refuses to employ it will be able to win.
No, the danger is that it isn’t a scam. Scams eventually get revealed, but the AI companies fix the places where it is failing. The difference between AI and human learning is that each person has to learn from scratch, while AIs build.
Don’t expect the government to take over. The government is going to depend on AI soon, and a lot of the government is going to be bought and paid for.
In the old days automation was cool because people displaced from their manual jobs can retrain into more intellectual jobs, like coding. But the coders are being replaced. Where are people going to go? Into healthcare? Into the arts? Into checking on the output of the AI.
If all this wealth being generated gets spread to those displaced, it might be okay. But we both know that’s not going to happen.
The Industrial Revolution is a good analogy. Dickensian slums were the direct outcome of it. Ned Ludd had a point, but he didn’t have the internet.
I bet the turning point, if there is one, will not be government doing anything but mobs of displaced workers destroying data centers.
Into ripping out all the AI and replacing it with real code after the crash. This “AI” is incapable of performing as advertised, it’s been compared to asbestos as something that’s toxic, everywhere, and will have to be expensively torn out and replaced.
As for government, don’t confuse a lack of interest with a lack of ability. A government could crush the wealthiest billionaire like an insect at any time they wanted; they just don’t care to do so. If a government changes their mind on the matter, well, lots of Russian oligarchs thought their money made them invincible until Putin had hem murdered.
The techbros might think that they can use AI to take over the world, but what would realistically happen is that their toys would be taken away from them.
This is true, but it seems to me that quite a lot of AI-produced material is either good enough or actually useful. I’m shouting from the rooftops because I think the minority that is hallucinated, flat-out wrong, and / or dangerous is an unacceptable risk, but others disagree, and of course if you’re only measuring outcomes by money, as is common, your tolerance for the risk of error goes way up.
Until everything falls apart because too many errors accumulate.
The push is remarkable. No one had to say “Get a TV or you’ll be left behind!” etc. The advantages were apparent. Either it’s a scam, or people are being pushed because the pushers (but not the pushed) are the only ones that will get the benefits.
A bit of both I think. It’s a scam, but prime targets of the scam are wealthy people and corporations who hate having employees with a passion, and see “AI” as their great chance to fulfill their dreams of a workplace without people. And being who they are, they have a great deal of power to force the scam on everyone else.
FWIW according to The Economist (podcast) anyway supply crunch of chips may soon be getting these companies wanting to, temporarily, throttle demand. The fabs can’t keep up.
"… So for example, Anthropic, which is a large AI model maker recently changed the terms of its service so that it dissuades a lot of its users from using capacity at peak times. OpenAI, which is another big AI company, shut down Sora, which is its video generation tool because it wanted to allocate its scarce computing resources towards more lucrative … Potentially that could then slow down adoption as well. And so there have been those that have called the supply crunch a quote unquote natural break on this reckless AI spending. That’s one view.
The other view is this could potentially slow down AI adoption in a very real way if it continues"
That’s like 90% of it. AI is the “shiny new thing”. Business (and people in general) have been chasing that since at least the 90s (when I started working). That next big once in a lifetime thing where “if only you bet a couple thousand dollars on it 20 years ago” you’d be so rich you’re family would never need to work again.
I think you overestimate the quality and difficulty of a lot of code. Sure there is some stuff beyond AI now, and the output is going to have to be carefully checked (which is also true for human written code) but there is a lot of stuff that AI can do a fine job on, increasing the productivity of the top people and eliminating those who they would give the carefully defined jobs to. This has been going on for ages. I taught sorting algorithms, but today no one writes one, you just use the existing package. I remember when assembly language programmers were sure their code was more efficient than that which came from compilers - then that became no longer true.
Before I retired the big project I ran and mostly wrote had too many moving parts for AI, but I also wrote lots of little utilities for my users which I could have used AI for today. Or with some training they might have written themselves.
Putin didn’t have the inconvenience of expensive elections. There is no government, there are people in government, and those that mortgage themselves to billionaires are not going to bite the hand that feeds them.
And I can’t believe I’m being more cynical than you.