Sam Altman Says 'Yes,' AI Is In a Bubble (theverge.com) 56
OpenAI CEO Sam Altman told reporters that AI investments have entered bubble territory. His remarks: "Are we in a phase where investors as a whole are overexcited about AI? My opinion is yes."
"When bubbles happen, smart people get overexcited about a kernel of truth. If you look at most of the bubbles in history, like the tech bubble, there was a real thing. Tech was really important. The internet was a really big deal. People got overexcited."
He added that he thinks it's "insane" that some AI startups with "three people and an idea" are receiving funding at such high valuations. "That's not rational behavior," Altman said. "Someone's gonna get burned there, I think. Someone is going to lose a phenomenal amount of money."
"When bubbles happen, smart people get overexcited about a kernel of truth. If you look at most of the bubbles in history, like the tech bubble, there was a real thing. Tech was really important. The internet was a really big deal. People got overexcited."
He added that he thinks it's "insane" that some AI startups with "three people and an idea" are receiving funding at such high valuations. "That's not rational behavior," Altman said. "Someone's gonna get burned there, I think. Someone is going to lose a phenomenal amount of money."
Slashdotters (Score:2)
Re: (Score:2)
S"omeone is going to lose a phenomenal amount of money."
Uh, d'uh!
That's the point.
It's working as expected.
Re:Slashdotters (Score:5, Insightful)
Sam says that investing in small AI startups is dumb.
Instead, investors should be giving him the money.
Re: (Score:2)
Exactly, the last line is pretty much saying "Do your homework and only invest with us".
Of course, if anyone did their homework, they would be keeping their AI investments to a minimum.
This "astronomist" keeps calling them "dark holes" (Score:4, Informative)
Re: (Score:2)
Tech bubbles all work the same way. People who get in early tend to make a lot of money. Employees of tech companies often do fairly well with little financial risk.
People who get in late, that maybe didn't even have the hype train on their radar until the mainstream media started reporting on it, those people are the ones that often get screwed. So the middle class retirement accounts are what props up this boom-bust grift cycle. Wealth is pumped from the middle of our economy up to the top floor.
disclaime
"ChatGPT, pretend to be the product" (Score:5, Insightful)
These will be the first to go.
Of course "go" in this case is get the money and run the hell away to Brazil or something.
Crash and burn, or rise and conquer (Score:5, Insightful)
"Someone's gonna get burned there, I think. Someone is going to lose a phenomenal amount of money."
Yep. Most of the AI companies will crash and burn; it's inevitable. A small percentage will go on to billions and billions of dollars, and ten years from now everybody will be saying "we shoulda invested in that AI firm when it was still affordable, we woulda been rich!"
Re: (Score:2)
Many are predicting that AI will be a "Winner takes all" market, like search, e-commerce, and social media.
One company will capture 80% of the market and 95% of the profits.
We don't know who the winner will be, but Sam's point is that it's likely to be one of the big companies, like his, not some tiny startup.
Summary of the summary: Sam says you should be giving your money to Sam.
Re: (Score:2)
The weird thing with this bubble is that the bubble startups are also dependent on the established players. Lots of startups do AI this or that but really just use an OpenAI or Claude API rather than do the hard part themselves. Regardless of the success or failure of these smaller companies, OpenAI gets to keep the money. They can ride the bubble until the end and still survive.
Re: (Score:2)
A lot of these smaller 3-12 person companies will develop some proprietary tech on top of (probably agnostic) state of the art models and rather than sell the product they just maintain it for their existing customers as a professional services company. Something that solves a difficult to solve problem, but has to be uniquely wired into each client's system differently, and then add/tweak features per customer.
Re: (Score:2)
Re: (Score:3)
Re: (Score:2)
Many of those small startups are not competitors
The small guys aren't Sam's competitors for products or services.
But they are his competitors for talent and VC funding.
Re: (Score:2)
I won't be surprised if AI turns into a game of wack-a-mole though where each "improvement" causes some new issue to surface. It could take decades to get right much like speech to text took since early 80's I think it was to even today works just ok. If I were negotiating a bi
Re: Crash and burn, or rise and conquer (Score:2)
Beta had superior quality but for years it had inferior time per tape, so even on a purely technical basis it was a mixed bag.
The superior tech won (Score:2)
Beta had superior quality but for years it had inferior time per tape, so even on a purely technical basis it was a mixed bag.
Exactly. Beta couldn't put a whole film on one tape, so the VHS, which COULD put a movie on one tape, won.
Nothing mysterious about it. For videotape, the most important criterion of "superior" means "able to put a movie on a casette". The superior technology, by this critetion, won.
Re: (Score:2)
Beta preceded VHS and was superior
VHS was good enough and was cheaper.
Re: (Score:2)
Re: (Score:2)
Sony beta was of higher picture quailty on a 1 hour tape.
VHS had poorer image quality, but could record a 2 hour program and later 4 and 6 hours at ever lower quality.
Consumers (and later video rental stores) wanted at least 2 hours of recording time as that was the median length of movies.
Sony made the superior product, but on too small a cassette and did not want to create a new larger cassette format.
Had Sony done that before 1980 with backward compatible adapter at the VHS price point and some expanded
Re: (Score:2)
Sam Altman Pivots to Manufacturing Hot Dog Suits (Score:2)
The maverick CEO promises an entire Silicon Valley techbro cult's worth of people trying to find the guy who did this by FY2026.
uh no (Score:5, Insightful)
"When bubbles happen, smart people get overexcited about a kernel of truth"
That's a funny way to spell "stupid and ignorant people" but it makes sense he'd spin it that way to avoid saying that his investors are idiots.
Re: (Score:1)
He's probably taking a cue from someone who might say something like this:
"The best people, the smartest people, are known. That kind of intelligence is needed to make America great. The economy and the stock market are all about being smart, having an incredible brain. I can tell you, there's a great brain, a tremendous brain. Everybody knows it, everybody says it. And you know, I went to the best schools, and was a very smart person. So we're going to keep bringing in smart people, the smartest people, to
The difference with AI is (Score:5, Interesting)
There aren't well defined limitations that are apparent yet (to everyone). Investors are assuming a linear growth of tech, LLMs->Reasoning models->Human AI-> Superhuman AI and all this will happen in just a few years with no roadblocks. Industries in growth/investment mode don't have to answer for R&D especially now, everyone is so worried about missing out on this investment they are going all in on ANYTHING AI related. But once the limitations set in there will be a pulling back. AI also only approximates human thinking and the training is done at a human level. Another problem is the smallest transistors use more energy than neurons on the order of roughly six orders of magnitude of energy to switch. At some point it's going to be cheaper to use a human than a super computer. Energy costs haven't set in.
Once realities set in there will be a bubble pop, it will be huge. There will be winners and losers. I think a lot of these investments will be scaled back. The real disruption is going to be AI 2.0 when people figure out ways to apply AI everywhere.
Re: (Score:2)
Waiting for see-through, I mean, AI-through peripherals.
Could really use an AI colored toaster and matching blend.
Re: (Score:2)
Textbook case for a human being cheaper than AI at loading a toaster.
Re:The difference with AI is (Score:5, Informative)
There are two types of AI stories that seem to come out almost every day at this point. The first kind is, "CEO says his company's products are amazing. If you don't use them, you'll get left behind." The second kind is, "Researchers find AI doesn't work nearly as well as companies claim." Both stories are repeated on a daily basis. Here are a few examples of the latter type, drawing only from stories posted to Slashdot in the last week.
https://slashdot.org/story/25/... [slashdot.org]
https://slashdot.org/story/25/... [slashdot.org]
https://news.slashdot.org/stor... [slashdot.org]
https://it.slashdot.org/story/... [slashdot.org]
This perhaps tells us something about where we are in the hype cycle.
Re: (Score:2)
The third type of stories, which Altman hates, is the "You can now run this open source model on your phone".
With all the data of the internet having been consumed twice for training, the only option now is efficiency, and more and more small-size precision models are coming out that are specialized for one use (Translation, programming,...) and are easier to run locally.
There will always be people who will just fork cash for an AI sub, but running one locally will also keep getting easier.
Re: (Score:2)
AI also only approximates human thinking and the training is done at a human level. Another problem is the smallest transistors use more energy than neurons on the order of roughly six orders of magnitude of energy to switch.
While the term 'neural network' is misleading, you should know better by now. Neural networks are absolutely nothing like brains. They do not learn, grow, or change as they're used and nothing they do is even remotely like thinking, let alone "human thinking". How much energy they need relative to one another is completely meaningless because the two things are not equivalent.
As for training, they are also not trained in any way that could be described as "at a human level". The way we train these things
Refund the money then (Score:1)
If I was an investor, I'd be mad at the guy I gave money to for admitting I invested my money in a whole lot of hot air.
Then again, if I was investing in this idiocy, I'd deserve everything I'd get...
Re: (Score:2)
Nah, he's not talking about investments in HIS company, those are totally not a bubble and will return HUGE gains! He's talking about the "teams of 3 people with an idea" getting billions of dollars in valuation (a point upon which I cannot disagree).
Re: (Score:2)
I think AI is overhyped and underhyped. People are way too focused on chatbots and think companies like AI will make AGI. That won't happen.
On the otherhand, people are missing the important things, like Alphafold2 solving protein folding. Alphafold3 making first steps for AI generated drugs, which could mean that we are few years away from immortality. Or AlphaEvolve improving several algorithms, as in automating scientific progressing.
Looking out for number 1... (Score:5, Insightful)
Because the insane thing is definitely the valuation of the random little guys with costs barely distinguishable from zero; rather than serious professionals who need to set some billions of dollars a year on fire just to keep the lights on.
I'm sure that this view of who is and who isn't bubble is entirely unrelated to which kind of AI startup Mr. Altman is running.
Why Bubbles show up (Score:5, Interesting)
Bubbles for Startups happen for one very important reason - Tax avoidance by the rich.
Say you have enough money to live your life totally safe. 13 million dollars is a good number (based on what American estate get no taxes). Then say you have 10 million more dollars and you want to invest it. But you hate the idea of paying taxes, particularly over small gains that may vanish 5 years later.
So you do what I call the lottery. You take your 10 million dollars and invest in 10 different long shot investments. If you have a good investment advisor, three years later you get:
3 investments that go broke.
3 investments that are flat.
3 investments that did well, up about 30% on average (about 10% per year).
1 block buster that is now worth 10 million.
Your total value is worth about 17 million (not including the 13 million you live on), a nice return for 3 years.
Now you sell everything except your blockbuster, you get 7 million back, with a cost basis of 9 million dollars. Go ahead and sell 20% of your block buster, you owe NO taxes, even though you have turned 10 million into 9 million in cash, plus 8 million of equity in a blockbuster.
What it comes down to is this - the US tax system rewards people that get both large gains and large losses, rather than moderate gains. Bubbles are LOVED by the wealthy - it gives them both things they want.
Re: (Score:2)
And sell the investments that are underwater a bit at a time to offset my capital gains.
And also have a line of credit for some fraction against all 10 of those example (such as an LAL). So you can borrow around $3m against your $10m of investment at a fixed rate (say 7.12% APR). And since it's credit, you don't pay taxes on it, it's not income. Now use that like a bridge loan to purchase investments, like real estate. And then to avoid eating that high interest you start loss harvesting your poor performin
Re: (Score:3)
1 block buster that is now worth 10 million.
I read this three times trying figure out how Blockbuster is worth 10 million [wikipedia.org]. It was a bad day to quit coffee.
Re: (Score:2)
This isn't a rich thing tho. It's actually more of a middle/upper-middle class thing.
It's called tax loss harvesting and is well known. You're definitely right about the lottery effect, but steady returns are (ignoring taxes) better because the jackpot might not happen, but in real life the highest returns ARE indeed lottery like (a very small # of companies account for most of and index's return).
The ultra wealth real advantage is they can borrow against their shares (which they sometimes are not allowed o
Humanity (Score:2)
Re: (Score:2)
He's right but not for the reasons you think he is (Score:5, Interesting)
You need training data to make llms work. And the bulk of the training data has been coming from publicly available sources like web forums crawled by bots.
Those forms are either starting to shut down the bots access or they are so full of AI slop they are useless for training.
That means in the near future the only people with access to training data are going to be major platform holders. Google, microsoft, Apple etc.
That means AI is a one-time Gold Rush that will quickly pick winners and losers and create a consolidated market with between two to three players.
As soon as the consolidation is done then yeah the bubble is going to burst. I'm not entirely sure how long but I would think within two or three years tops with how fast things go now. I wouldn't expect more than five but I really am just guessing.
Re: (Score:2)
Re: (Score:2)
AI == 3D_TV (Score:2)
We'll be witnessing a recurring surge of "AI" just like it happens with 3D TVs.
Sure, tech improves each time but the real deal is still on the horizon.
Or to put it in a car analogy, flying cars.
Re: (Score:2)
There's a big difference between AI and 3D TVs. The difference is, everybody *wants* to use AI.
Re: (Score:2)
He's just one of many conmen that plague us right now.
"Bubbles" always happen for every nascent field (Score:2)
The idea that a bunch of investment money rushes into a bunch of companies in a new promising field and that many/most of those companies will eventually falter is the normal evolution of every single big and most small things that have happened in tech over the last hundred years. This is normal and good and to be expected. Of course, a lot of investors will lose a lot of money, a few will gain a fortune, but at least from the perspective of society, a big win results.
SAM ALTMAN GENIUS - NO (Score:1)
There are geniuses among us. Sam Altman is not that guy.
I could go on but I'll save you the brain cells.
He's a PR guy. No more. No less. What he says is PR. His company has no product. AI is not a thing.
But hey, don't go investing in it. It's not an investment. It's a gamble on the stupidity of others.
You heard it here on slashdot on august 15th. Later on when you've lost your "investment" I'm sorry. Sorry you didn't listen.
Sorry you don't know what "investment" vs "gamble" is. Sorry you got take