![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
thewayneI'm just going to copy the Slashdot summary, then comment on it: Fast Company ran a contrarian take about AI from entrepreneur/thought leader Faisal Hoque, who argues there's three AI bubbles.
The first is a classic speculative bubble, with asset prices soaring above their fundamental values (like the 17th century's Dutch "tulip mania"). "The chances of this not being a bubble are between slim and none..."
Second, AI is also arguably in what we might call an infrastructure bubble, with huge amounts being invested in infrastructure without any certainty that it will be used at full capacity in the future. This happened multiple times in the later 1800s, as railroad investors built thousands of miles of unneeded track to serve future demand that never materialized. More recently, it happened in the late '90s with the rollout of huge amount of fiber optic cable in anticipation of internet traffic demand that didn't turn up until decades later. Companies are pouring billions into GPUs, power systems, and cooling infrastructure, betting that demand will eventually justify the capacity. McKinsey analysts talk of a $7 trillion "race to scale data centers" for AI, and just eight projects in 2025 already represent commitments of over $1 trillion in AI infrastructure investment. Will this be like the railroad booms and busts of the late 1800s? It is impossible to say with any kind of certainty, but it is not unreasonable to think so.
Third, AI is certainly in a hype bubble, which is where the promise claimed for a new technology exceeds reality, and the discussion around that technology becomes increasingly detached from likely future outcomes. Remember the hype around NFTs? That was a classic hype bubble. And AI has been in a similar moment for a while. All kinds of media — social, print, and web — are filled with AI-related content, while AI boosterism has been the mood music of the corporate world for the last few years. Meanwhile, a recent MIT study reported that 95% of AI pilot projects fail to generate any returns at all.
But the article ultimately argues there's lessons in the 1990s dotcom boom: that "a thing can be hyped beyond its actual capabilities while still being important... When valuations correct — and they will — the same pattern will emerge: companies that focus on solving real problems with available technology will extract value before, during, and after the crash." The winners will be companies with systematic approaches to extracting value — adopting mixed portfolios with different time horizons and risk levels, while recognizing organizational friction points for a purposeful (and holistic) integration.
"The louder the bubble talk, the more space opens for those willing to take a methodical approach to building value."
The first bubble is obvious. Huge amounts of money is being 'invested' in AI/LLMs and the returns have been dubious and amusing, and sometimes lethal. Children and teens taking their own lives, a formerly well-behaved autistic child becoming violent, etc. The valuation of Tesla going up while its sales sales plunge is always an amusing example. The infrastructure bubble is tragic: coal and offline nuclear power plants are being planned to power data centers exclusively for these things, and along with them are their water requirements. And that is a really big problem with increasing climate change. I read an article that I'll post if I can find it that said that each simple AI query is the equivalent of the use of a small bottle of water. The ecological cost is really quite, quite staggering. The eco cost of bitcoin and its kin is trivial compared to this.
The third bubble is interesting. They've demonstrated that LLMs can do some very cool things when tasked into specific purposes and trained in specific bodies of knowledge, like researching new antibiotics or metal alloys with new properties that are needed.
I think the thing that I'm the most curios about is when the corrections/collapses will start taking place. Considering the valuations involved, the financial quake will make the Dot Com crash look like the merest tremor.
The author, Faisal Hoque, is a lot more optimistic about AI than I. He compares its development to such as Amazon and Google during the Dot Com era of the 90s. They had very long-term development timelines ('Moon Shots') that they were quietly pursuing that achieved their long-term survival. And while not all current AI companies are going to achieve those and remain largely in their current form, some may. He talks about Pets.com burning through $300mil before collapsing, which we now see as a trivially small amount of money in today's tech market.
Curious times. We shall see how things shake out.
https://www.fastcompany.com/91400857/there-isnt-an-ai-bubble-there-are-three-ai-bu
https://slashdot.org/story/25/09/20/1847246/there-isnt-an-ai-bubble---there-are-three
The first is a classic speculative bubble, with asset prices soaring above their fundamental values (like the 17th century's Dutch "tulip mania"). "The chances of this not being a bubble are between slim and none..."
Second, AI is also arguably in what we might call an infrastructure bubble, with huge amounts being invested in infrastructure without any certainty that it will be used at full capacity in the future. This happened multiple times in the later 1800s, as railroad investors built thousands of miles of unneeded track to serve future demand that never materialized. More recently, it happened in the late '90s with the rollout of huge amount of fiber optic cable in anticipation of internet traffic demand that didn't turn up until decades later. Companies are pouring billions into GPUs, power systems, and cooling infrastructure, betting that demand will eventually justify the capacity. McKinsey analysts talk of a $7 trillion "race to scale data centers" for AI, and just eight projects in 2025 already represent commitments of over $1 trillion in AI infrastructure investment. Will this be like the railroad booms and busts of the late 1800s? It is impossible to say with any kind of certainty, but it is not unreasonable to think so.
Third, AI is certainly in a hype bubble, which is where the promise claimed for a new technology exceeds reality, and the discussion around that technology becomes increasingly detached from likely future outcomes. Remember the hype around NFTs? That was a classic hype bubble. And AI has been in a similar moment for a while. All kinds of media — social, print, and web — are filled with AI-related content, while AI boosterism has been the mood music of the corporate world for the last few years. Meanwhile, a recent MIT study reported that 95% of AI pilot projects fail to generate any returns at all.
But the article ultimately argues there's lessons in the 1990s dotcom boom: that "a thing can be hyped beyond its actual capabilities while still being important... When valuations correct — and they will — the same pattern will emerge: companies that focus on solving real problems with available technology will extract value before, during, and after the crash." The winners will be companies with systematic approaches to extracting value — adopting mixed portfolios with different time horizons and risk levels, while recognizing organizational friction points for a purposeful (and holistic) integration.
"The louder the bubble talk, the more space opens for those willing to take a methodical approach to building value."
The first bubble is obvious. Huge amounts of money is being 'invested' in AI/LLMs and the returns have been dubious and amusing, and sometimes lethal. Children and teens taking their own lives, a formerly well-behaved autistic child becoming violent, etc. The valuation of Tesla going up while its sales sales plunge is always an amusing example. The infrastructure bubble is tragic: coal and offline nuclear power plants are being planned to power data centers exclusively for these things, and along with them are their water requirements. And that is a really big problem with increasing climate change. I read an article that I'll post if I can find it that said that each simple AI query is the equivalent of the use of a small bottle of water. The ecological cost is really quite, quite staggering. The eco cost of bitcoin and its kin is trivial compared to this.
The third bubble is interesting. They've demonstrated that LLMs can do some very cool things when tasked into specific purposes and trained in specific bodies of knowledge, like researching new antibiotics or metal alloys with new properties that are needed.
I think the thing that I'm the most curios about is when the corrections/collapses will start taking place. Considering the valuations involved, the financial quake will make the Dot Com crash look like the merest tremor.
The author, Faisal Hoque, is a lot more optimistic about AI than I. He compares its development to such as Amazon and Google during the Dot Com era of the 90s. They had very long-term development timelines ('Moon Shots') that they were quietly pursuing that achieved their long-term survival. And while not all current AI companies are going to achieve those and remain largely in their current form, some may. He talks about Pets.com burning through $300mil before collapsing, which we now see as a trivially small amount of money in today's tech market.
Curious times. We shall see how things shake out.
https://www.fastcompany.com/91400857/there-isnt-an-ai-bubble-there-are-three-ai-bu
https://slashdot.org/story/25/09/20/1847246/there-isnt-an-ai-bubble---there-are-three
no subject
Date: 2025-09-21 07:59 pm (UTC)Interesting and well-put. I expect much the same, but for financial matters, I'm aware I'm a complete idiot and will refrain from trying to so invest.
no subject
Date: 2025-09-21 08:04 pm (UTC)I would have liked to have invested in Nvidia about three years when my dad passed away, just a few thousand, but my sister was very slow about kicking my brother out of the house and selling the place. I missed the sweet spot in their growth, and while they will continue to grow for a while, I think said growth is slowing. I really don't feel comfortable about directly investing in an "AI" company.
no subject
Date: 2025-09-21 08:52 pm (UTC)no subject
Date: 2025-09-21 08:53 pm (UTC)no subject
Date: 2025-09-22 12:59 am (UTC)There's a lot of popular opinion that when a lot of the AI companies go bust that their server farms will go cheap. Plus, as rapidly as NVidia is evolving their GPUs, those things do get swapped out on a pretty fast cycle. The problem is, though technically they are GPUs, they don't have video out ports. As far as video ram is concerned, I don't know anything specifically about that in regards to AI processing.
no subject
Date: 2025-09-22 07:18 am (UTC)no subject
Date: 2025-09-22 04:01 pm (UTC)Yeah, for that purpose they work fine. Just not for gaming video cards, as some people think they can be used for. I watched a video, probably over a year or so ago now - guy acquired a true blockchain GPU. He wanted to run video benchmarks against it, but it had no video output or video drivers. He ended up running something like Monero generation on it versus some high-end video card. Surprise, it performed better than the video card for that task. I think my desktop runs a 4070, my laptop (Asus TUF Gaming F16) runs an Ultra 9 with an RTX 5070, but it also has an NMU, I think they call it. I have no idea what to do with that aspect. But the Ultra 9 does do multithreading with old games much better than the i7 does: I run 16 instances of Lord of the Rings Online - it pegs the i7, but runs about 40% of the capacity of the U9.
no subject
Date: 2025-09-22 07:45 pm (UTC)no subject
Date: 2025-09-22 07:59 pm (UTC)I run a band. :-) It really hammers an i7, and with 64 gig of ram in that box eats 52ish gig. My laptop with the U9 and 32 gig it eats 25ish and runs about 40%, as I said. The video card works kinda hard as it has only 8 gig.
no subject
Date: 2025-09-22 08:05 pm (UTC)Half an hour of a Yule holiday performance by my band three years ago. https://www.youtube.com/watch?v=Pubutws9Cx4&list=RDPubutws9Cx4&start_radio=1
no subject
Date: 2025-09-22 10:18 pm (UTC)no subject
Date: 2025-09-22 11:26 pm (UTC)I'll dig up a link to Brandywine Theater Company's production of The Wall. As I speak, we're doing rehearsals for Fiddler On The Roof, to be performed in November. We're also doing two performances of Rocky Horror Picture Show for Halloween! I'm doing the rehearsal band occasionally for Fiddler and will be the performance band for Rocky. Someone else is composing the music for the Brandywine productions, they're much better than me. And the people who make this game, Standing Stone Games (SSG), are pretty gobsmacked at what BTC is able to do! Some of their game moderators attend our performances. :-)
no subject
Date: 2025-09-23 03:21 am (UTC)This is an hour of Queen from an annual Monsters of the Rock show. Lots of fun! https://www.youtube.com/watch?v=2D8L6yYvy20&list=RD2D8L6yYvy20
no subject
Date: 2025-10-07 01:37 am (UTC)I'm part way through listening to the entire holiday festival (& watching a fair bit of it; I love how people wind up line dancing, it's glorious) & just got to my favorite version of my favorite Christmas song ever, & now I'm picturing the entire band set up outside the Black Gates with the armies of the west behind them & playing that right at Sauron. Good idea? Nah. But BEST FUCKING IDEA? HELL YEAH
... I should have known better than to try to eat through Mordor Christmas Carol
if someone else is composing the music for Brandywine, does that mean you're doing the orchestrations for your band? because if so, DAMN, sir
plus you're doing ENTIRE ASS PLAYS, THAT'S SO COOL
no subject
Date: 2025-10-07 03:30 am (UTC)I have plans for other music videos with custom lyrics. With what I'm learning from Brandywine Theater Co., I think I can do much better! Springfest Oddity/Sgt Tom was very vaguely scripted and shot, there's lots to improve upon. It's certainly possible to get in front of the Black Gates, but it's a bit dangerous. You'd need a level 150 patrol group to keep the band safe while they're performing. Which could make for interesting excerpts.
World of Warcraft had an easier way of doing things. They had a server that you could create instances in, then use a software package that let you program avatar actions in and render that into a video. You could do some pretty awesome stuff that way. We have to work harder, but it's worth it.
no subject
Date: 2025-09-21 11:28 pm (UTC)AI is also useful in a sense for high school and college students who don’t want to write their own papers, but this seems to be socially the opposite of useful: young people will fail to develop the skills of doing research (when appropriate), making a well-reasoned case, and finding the right words to set forth that case. Granted, many people never got very good at this even before AI, but students might have to acquire these skills to some extent.
As to children and teens taking their own lives, there may be a moral panic about that. I don’t know how many cases of suicide or other major harm there have been due to AI chat, and how many young people have found AI to be a source of help or a voice of sanity when the human beings around them were nasty, or didn’t understand them.
no subject
Date: 2025-09-22 01:04 am (UTC)So many technologies are two-edged swords, but that's true of many sciences. My wife got into astronomy pretty much specifically as it doesn't have military applications. It's been demonstrated that it's far too easy to bypass the safeguards and guard rails that chatbot makers put in place, and this is a very serious problem. Once these things are running, I have a feeling that it's almost impossible to truly understand the code of how they work. I believe Facebook engineers have admitted that they don't know exactly how the overall platform works: they can make changes and tweaks, but they don't know the entire system, and that's not a full-fledged LLM.
no subject
Date: 2025-09-22 12:55 am (UTC)Call me old, but I just as soon stay away from AI. :o
Hugs, Jon
no subject
Date: 2025-09-22 01:08 am (UTC)TBH, I get some good answers from AI, but I use it in very limited ways in subjects that I know. And I don't use them over long periods of time when they're likely to start hallucinating. I'm just way too suspicious to converse with them and treat them like a buddy. I'll never trust them for that.
no subject
Date: 2025-09-23 05:28 am (UTC)Nope and Nope.
LOL!!!!!!!!!!!!!!!!!!!!!!!
no subject
Date: 2025-09-23 05:52 am (UTC)You can do an -AI at the end of your search terms in Google to turn off the AI suggestions.
no subject
Date: 2025-09-23 05:58 am (UTC)no subject
Date: 2025-09-22 02:34 am (UTC)For as many years as possible, it will get sucked out by engineers and CTOs and manufacturers, and blown on real-world goods like property, vehicles, and fine dining, or squirreled away somehow, and when there are no more houses of cards to run between because they've all collapsed, we will turn our attention to the industrious winners with their long-term plans.
I've said this before, but I can't remember where. It bears repeating:
The state of the art does not suddenly appear everywhere at once. Instead it advances like the crest of a wave. The wave of advancement always has a very long slope, leading down to the people in the trough who are struggling right now to just to get food and shelter. It rushes around the earth and we like to think the wave only lifts people up, but many people just get churned before it instead. The inequality, the exploitation, the fleecing of the unwary ... that's what we need to pay attention to. Even though - let's face it - it's less interesting.
no subject
Date: 2025-09-22 04:07 am (UTC)I can just picture people barely making it getting sucked in by advertising that they 'must!' buy a new AI-enabled TV. I just saw where Samsung, with their LCD-screen-equipped refrigerators are now going to be pushing ads onto said screens. It is such a shame that with all the tech we have and all the money behind it, all that's going on is the push to ever-increase the money. Almost zero effort to improve the standard of living of the lowest class.
no subject
Date: 2025-09-22 04:47 am (UTC)no subject
Date: 2025-09-22 03:51 pm (UTC)Ah, but there's easy financing!
no subject
Date: 2025-09-22 07:46 pm (UTC)no subject
Date: 2025-09-22 08:00 pm (UTC)Oh, yeah! NEVER in my house! I made sure when I last bought a washer to avoid an internet connection. Apparently it does have WiFi but doesn't require an internet connection to work, so I've never looked into it.
no subject
Date: 2025-09-27 09:10 pm (UTC)It's going to be good for those who are professional debuggers and rewriters of code, certainly, but I'm not sure it's going to go over well with the people who have been replaced by a sub-par chatbot, or who are playing the part of the Mechanical Turk to those chatbots.
no subject
Date: 2025-09-28 01:15 am (UTC)I saw an article today about the energy cost of LLM-generated video. It doesn't scale linearly. "...a six-second AI video clip consumes four times as much energy as a three-second clip." Yeah, I agree. Data centers are the only recoverable asset in this tri-boom. In other reporting, the adoption rate of LLMs in corporate America seems to be slowing, which can indicate saturation or acknowledgement that it isn't doing much to improve productivity or profits. With the growth of 'vibe-code fixer shops', I'm suspecting more of the latter.
no subject
Date: 2025-09-28 05:37 am (UTC)Also, O^2 or worse time for running tasks also makes sense, since each pixel of reach frame has to go through the entire guessing and comparing procedure.