Maybe it’s like the dotcom bubble: there is genuinely useful tech that has recently emerged, but too many companies are trying to jump on the bandwagon.
LLMs do seem genuinely useful to me, but of course they have limitations.
We need to stop viewing it as artificial intelligence. The parts that are worth money are just more advanced versions of machine learning.
Being able to assimilate a few dozen textbooks and pass a bar exam is a neat parlor trick, but it is still just a parlor trick.
Unfortunately probably the biggest thing to come out of it will be the marketing aspect. If they spend enough money to train small models on our wants and likes it will give them tremendous amounts of return.
The key to using it in a financially successful manner is finding problems that fit the bill. Training costs are fairly high, quality content generation is also rather expensive. There are sticky problems around training it from non-free data. Whatever you’re going to use it for either needs to have a significant enough advantage to make the cost of training /data worth it.
I still think we’re eventually going to see education rise. The existing tools for small content generation adobe’s use of it to fill in small areas is leaps and bounds better than the old content aware patches. We’ve been using it for ages for speech recognition and speech generation. From there it’s relatively good at helper roles. Minor application development, copy editing, maybe some VFX generation eventually. Things where you still need a talented individual to oversee it but it can help lessen the workload.
There are lots of places where it’s being used where I think it’s a particularly poor fit. AI help desk chatbots, IVR scenarios, It says brain dead as the original phone trees and flow charts that we’ve been following for decades.
We’re hitting logarithmic scaling with the model trainings. GPT-5 is going to cost 10x more than GPT-4 to train, but are people going to pay $200 / month for the gpt-5 subscription?
Businesses might pay big money for LLMs to do specific tasks. And if chip makers invest more in NPUs then maybe LLMs will become cheaper to train. But I am just speculating because I don’t have any special knowledge of this area whatsoever.
I am old enough to remember when the CEO of Nortel Networks got crucified by Wall Street for saying in a press conference that the telecom/internet/carrier boom was a bubble, and the fundamentals weren’t there (who is going to pay for long distance anymore when calls are free over the internet? where are the carriers-- Nortel’s customers-- going to get their income from?). And 4 years later Nortel ceased to exist. Cisco crashed too, though had enough TCP/IP router biz and enterprise sales to keep them alive even until today.
This all reminds me of the late 1990s internet bubble rather than the more recent crypto bubble. We’ll all still be using ML models for all kinds of things more or less forever from now on, but it won’t be this idiotic hype cycle and overvaluation anymore after the crash.
Shit, crypto isn’t going anywhere either, it’s a permanent fixture now, Wall Street bought into it and you can buy crypto ETFs from your stockbroker. We just don’t have to listen to hype about it anymore.
We just don’t have to listen to the hype about it anymore.
True, it’s now in most circles just been mixed in as a commodity to trade on. Though I wish everyone would get that. There’s still plenty of idiots with .eth usernames who think there’s some new boon to be made. The only “apps” built on crypto networks were and are purely for trading crypto, I’ve never seen any real tangible benefit to society come out of it. It’s still used plenty for money laundering, but regulators are (slowly) catching up. And it’s still by far the easiest way to demonstrate what happens to unregulated markets.
Crypto has been turned into gold by wallstreet, they bought up enough of it to jot be completely exposed, it’s supply is extremely limited and will run out. Putting your money into it is no different than putting it into gold, you might catch a good moment and buy in low and get some return, but most wont.
Wow, a CEO who doesn’t buy into the hype? That’s astonishing.
I, for one, cannot wait for the bubble to burst so we can get back to some sense of sanity.
Edit>> Though if Baidu is investing in AI like all the rest, then maybe they just think they’ll be immune — in which case I’m sad again that I haven’t yet come across a CEO who calls bullshit on this nonsense.
AI will have its uses, and it has practical use cases such as helping people to walk or to speak or to translate in real time, etc. But we’re decades away from what all these CEOs seem to think they’re going to cash in on now. And it’ll be fun on some level watching them all be wrong.
10 to 30? Yeah I think it might be a lot longer than that.
Somehow everyone keeps glossing over the fact that you have to have enormous amounts of highly curated data to feed the trainer in order to develop a model.
Curating data for general purposes is incredibly difficult. The big medical research universities have been working on it for at least a decade, and the tools they have developed, while cool, are only useful as tools too a doctor that has learned how to use them. They can speed diagnostics up, they can improve patient outcome. But they cannot replace anything in the medical setting.
The AI we have is like fancy signal processing at best
Creating a specialized neural net to perform a specific function is cool. Slapping GPT into customer support because you like money is horse shit and I hope your company collapses. But yeah you’re right. Blockchain was a solution with basically no problems to fix. Neural nets are a tool that can do a ton of things, but everyone is content to use them as a hammer.
I’m not even understanding what AI is at this point because there’s no delineation between moderately sophisticated algorithms and things that are orders of magnitude more complex.
I mean, if something like multisampling came out today we’d all know how it’d be marketed
AI is a ridiculous broad term these days. Everybody had been slapping the label on anything. It’s kinda like saying “transportation” and it means anything between babies crawling up to wrap drive and teleportation.
Yes. But companies bought into AI way more than they bought into crypto though, in many outlandish and stupid ways. And many AI companies sell it in ways they shouldn’t.
As a major locally-hosted AI proponent, aka a kind of AI fan, absolutely. I’d wager it’s even worse than crypto, and I hate crypto.
What I’m kinda hoping happens is that bitnet takes off in the next few months/years, and that running a very smart model on a phone or desktop takes milliwatts… Who’s gonna buy into Sam Altman $7 trillion cloud scheme to burn the Earth when anyone can run models offline on their phones, instead of hitting APIs running on multi-kilowatt servers?
And ironically it may be a Chinese company like Alibaba that pops the bubble, lol.
As someone who follows the startup space (and is thinking of starting their own, non-AI driven startup), the issue is all of the easily solvable problems have already been solved. The only thing that shakes up the tree is when new tech comes along and makes some of the old problems easy to solve.
So take a look at crypto - If you wanted to make a tip bot on Telegram, before crypto that was really hard. You needed to register with something like PayPal, have the recipient register with PayPal, etc etc etc. After crypto it was “Hey this person sent you 5$, use this private key if you want to recover it” (btw I made this service and it was used a lot).
Now look at AI - Imagine making a service that detects CSAM before AI took off. As an aside, I did NOT make this service, but I know a group of people who did. Imagine trying to make this without the AI boom - you’d need millions of images for training data, a PhD in machine learning, and so much more. Now, anyone can make it in their basement.
The point is, investors KNOW the bubble is a bubble and that it will pop. It doesn’t matter though. They’re looking for people who will solve problems that previously cost 1bln to solve with only 1mln of funding. If even 1% of their companies pay off, they make a profit.
problem is, the amount of soap(money) that goes around to make the bubbles keeps shrinking because the bubbles are siphoning it away from the consumers.
I wonder what happens when there’s no more soap left to go around?
I think less restrictive AI that are free, like Venice AI (you can ask it pretty much anything and it will not stop you) will be around for longer than ones that went with restrictive subscription models, and that eventually those other ones will become niche.
New technology always propagates further the freer it is to use and experiment with, and ChatGPT and OpenAI are quite restrictive and money hungry.
You are not logged in. However you can subscribe from another Fediverse account, for example Lemmy or Mastodon. To do this, paste the following into the search field of your instance: !technology@lemmy.world
This is a most excellent place for technology news and articles.
Maybe it’s like the dotcom bubble: there is genuinely useful tech that has recently emerged, but too many companies are trying to jump on the bandwagon.
LLMs do seem genuinely useful to me, but of course they have limitations.
We need to stop viewing it as artificial intelligence. The parts that are worth money are just more advanced versions of machine learning.
Being able to assimilate a few dozen textbooks and pass a bar exam is a neat parlor trick, but it is still just a parlor trick.
Unfortunately probably the biggest thing to come out of it will be the marketing aspect. If they spend enough money to train small models on our wants and likes it will give them tremendous amounts of return.
The key to using it in a financially successful manner is finding problems that fit the bill. Training costs are fairly high, quality content generation is also rather expensive. There are sticky problems around training it from non-free data. Whatever you’re going to use it for either needs to have a significant enough advantage to make the cost of training /data worth it.
I still think we’re eventually going to see education rise. The existing tools for small content generation adobe’s use of it to fill in small areas is leaps and bounds better than the old content aware patches. We’ve been using it for ages for speech recognition and speech generation. From there it’s relatively good at helper roles. Minor application development, copy editing, maybe some VFX generation eventually. Things where you still need a talented individual to oversee it but it can help lessen the workload.
There are lots of places where it’s being used where I think it’s a particularly poor fit. AI help desk chatbots, IVR scenarios, It says brain dead as the original phone trees and flow charts that we’ve been following for decades.
We’re hitting logarithmic scaling with the model trainings. GPT-5 is going to cost 10x more than GPT-4 to train, but are people going to pay $200 / month for the gpt-5 subscription?
Businesses might pay big money for LLMs to do specific tasks. And if chip makers invest more in NPUs then maybe LLMs will become cheaper to train. But I am just speculating because I don’t have any special knowledge of this area whatsoever.
I am old enough to remember when the CEO of Nortel Networks got crucified by Wall Street for saying in a press conference that the telecom/internet/carrier boom was a bubble, and the fundamentals weren’t there (who is going to pay for long distance anymore when calls are free over the internet? where are the carriers-- Nortel’s customers-- going to get their income from?). And 4 years later Nortel ceased to exist. Cisco crashed too, though had enough TCP/IP router biz and enterprise sales to keep them alive even until today.
This all reminds me of the late 1990s internet bubble rather than the more recent crypto bubble. We’ll all still be using ML models for all kinds of things more or less forever from now on, but it won’t be this idiotic hype cycle and overvaluation anymore after the crash.
Shit, crypto isn’t going anywhere either, it’s a permanent fixture now, Wall Street bought into it and you can buy crypto ETFs from your stockbroker. We just don’t have to listen to hype about it anymore.
True, it’s now in most circles just been mixed in as a commodity to trade on. Though I wish everyone would get that. There’s still plenty of idiots with .eth usernames who think there’s some new boon to be made. The only “apps” built on crypto networks were and are purely for trading crypto, I’ve never seen any real tangible benefit to society come out of it. It’s still used plenty for money laundering, but regulators are (slowly) catching up. And it’s still by far the easiest way to demonstrate what happens to unregulated markets.
https://www.web3isgoinggreat.com/
Crypto has been turned into gold by wallstreet, they bought up enough of it to jot be completely exposed, it’s supply is extremely limited and will run out. Putting your money into it is no different than putting it into gold, you might catch a good moment and buy in low and get some return, but most wont.
The supply is absolutely more like unlimited lol.
Not enough btc? Make lite coin! Etc etc etc
No one cares about lite coin though which defeats the purpose.
Etherium bla bla bla there are tons of them thousands. There us no shortage lol.
Silver, bronze etc, are you being dense on purpose? Though your seeming affinity for crypto implies you are simply dense.
Wow, a CEO who doesn’t buy into the hype? That’s astonishing.
I, for one, cannot wait for the bubble to burst so we can get back to some sense of sanity.
Edit>> Though if Baidu is investing in AI like all the rest, then maybe they just think they’ll be immune — in which case I’m sad again that I haven’t yet come across a CEO who calls bullshit on this nonsense.
AI will have its uses, and it has practical use cases such as helping people to walk or to speak or to translate in real time, etc. But we’re decades away from what all these CEOs seem to think they’re going to cash in on now. And it’ll be fun on some level watching them all be wrong.
10 to 30? Yeah I think it might be a lot longer than that.
Somehow everyone keeps glossing over the fact that you have to have enormous amounts of highly curated data to feed the trainer in order to develop a model.
Curating data for general purposes is incredibly difficult. The big medical research universities have been working on it for at least a decade, and the tools they have developed, while cool, are only useful as tools too a doctor that has learned how to use them. They can speed diagnostics up, they can improve patient outcome. But they cannot replace anything in the medical setting.
The AI we have is like fancy signal processing at best
No bubble has deserved to pop as much as AI deserves to
Blockchain and crypto were worse. „AI” has some actual use even if it’s way overblown.
I’m glad you didn’t say NFTs because my Bored Ape will regain and triple its value any day now!
Creating a specialized neural net to perform a specific function is cool. Slapping GPT into customer support because you like money is horse shit and I hope your company collapses. But yeah you’re right. Blockchain was a solution with basically no problems to fix. Neural nets are a tool that can do a ton of things, but everyone is content to use them as a hammer.
I’m not even understanding what AI is at this point because there’s no delineation between moderately sophisticated algorithms and things that are orders of magnitude more complex.
I mean, if something like multisampling came out today we’d all know how it’d be marketed
AI is a ridiculous broad term these days. Everybody had been slapping the label on anything. It’s kinda like saying “transportation” and it means anything between babies crawling up to wrap drive and teleportation.
Technically speaking how I differentiate it is:
Yes. But companies bought into AI way more than they bought into crypto though, in many outlandish and stupid ways. And many AI companies sell it in ways they shouldn’t.
Try Venice Ai, free to use, won’t try to censor your topics. Still just a chat bot though (although I think it does image generation too).
And they will ALL deserve it.
As a major locally-hosted AI proponent, aka a kind of AI fan, absolutely. I’d wager it’s even worse than crypto, and I hate crypto.
What I’m kinda hoping happens is that bitnet takes off in the next few months/years, and that running a very smart model on a phone or desktop takes milliwatts… Who’s gonna buy into Sam Altman $7 trillion cloud scheme to burn the Earth when anyone can run models offline on their phones, instead of hitting APIs running on multi-kilowatt servers?
And ironically it may be a Chinese company like Alibaba that pops the bubble, lol.
If bitnet takes off, that’s very good news for everyone.
The problem isn’t AI, it’s AI that’s so intensive to host that only corporations with big datacenters can do it.
The fuck is bitnet
As someone who follows the startup space (and is thinking of starting their own, non-AI driven startup), the issue is all of the easily solvable problems have already been solved. The only thing that shakes up the tree is when new tech comes along and makes some of the old problems easy to solve.
So take a look at crypto - If you wanted to make a tip bot on Telegram, before crypto that was really hard. You needed to register with something like PayPal, have the recipient register with PayPal, etc etc etc. After crypto it was “Hey this person sent you 5$, use this private key if you want to recover it” (btw I made this service and it was used a lot).
Now look at AI - Imagine making a service that detects CSAM before AI took off. As an aside, I did NOT make this service, but I know a group of people who did. Imagine trying to make this without the AI boom - you’d need millions of images for training data, a PhD in machine learning, and so much more. Now, anyone can make it in their basement.
The point is, investors KNOW the bubble is a bubble and that it will pop. It doesn’t matter though. They’re looking for people who will solve problems that previously cost 1bln to solve with only 1mln of funding. If even 1% of their companies pay off, they make a profit.
bubble after bubble after bubble after…
problem is, the amount of soap(money) that goes around to make the bubbles keeps shrinking because the bubbles are siphoning it away from the consumers.
I wonder what happens when there’s no more soap left to go around?
Checks to see if Baidu is doing AI…yes, they are. How shocking.
YUP
Tech makes bubbles because people think it’s magic.
I think less restrictive AI that are free, like Venice AI (you can ask it pretty much anything and it will not stop you) will be around for longer than ones that went with restrictive subscription models, and that eventually those other ones will become niche.
New technology always propagates further the freer it is to use and experiment with, and ChatGPT and OpenAI are quite restrictive and money hungry.