this post was submitted on 02 Jun 2024
193 points (91.8% liked)
Technology
59414 readers
3554 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
It's a hype bubble. AI had been around for a while and will continue to be. The problem is specifically Large Language Models. They've been trained to SOUND human, but not to actually use that ability for anything more useful than small talk and bullshit. However because it SOUNDS charasmatic, and that is interesting to people, companies have started cramming it into everything they can think of to impress shareholders.
Shareholders are a collective group of people who are, on average, really more psychologically similar to crows than other humans - they like shiny things, have a mob mentality, and can only use the most basic of tools available, in their case usually money. New things presented in a flashy way by a charasmatic individual are most attractive to them, and they will seldom do any research beyond superficial first impressions. Any research they actually do generally skews towards confirmation bias.
This leads to an unfortunate feature of capitalism, which is the absolute need to make the numbers go up. To impress their shareholders, companies have to jangle keys in front of their faces. So whenever The Hip New Things comes along, it's all buzzwords and bullshit as they try and find any feasible way to cram it into their product. If they could make Smart Corn 2.0 powered by Chat GPT they would, and sell it for three times as much in the same produce isle as normal corn. And then your corn would tell you this great recipe if knows where the sauce is made with a battery acid base.
In most recent memory, this exact scenario played out with NFTs. When the NFT market collapsed as was inevitable, the corporations who swore it would supercharge their sales all quietly pretended it never happened. Soon something new was jangled in front of the shareholders and everybody forgot about them.
Now that generative AI is proving itself to just be a really convincing bullshitter, it's only a matter of time until it either dies and quietly slinks away or mutates into the next New Things and the cycle repeats. Like a pandemic of greed and stupidity. Maybe they'll figure out how to teach Chat GPT how to check and cite verified sources and make it actually do what they currently claim it does.
I guess it depends on if they can make it shiny enough to impress the crows.
Hey this comparison is unfair to crows, they're way smarter and more empathetic than wealthy tech industry investors.
I think we're in an ai bubble because Nvidia is way over valued and I agree with you that often people flock to shiny new things and many people are taking risk with the hope of making it big...and many will get left holding the bag.
But how do you go from NFTs, which never had widespread market support, to the market pumping a trillion dollars into Nvidia alone? This makes no sense. And to down play this as "just a bullshitter" leads me to believe you have like zero real world experience with this. I use copilot for coding and it's been a boost to productivity for me, and I'm a seasoned vet. Even the ai search results, which many times have left me scratching my head, have been a net benefit to me in time savings.
And this is all still pretty new.
While I think there it is over hyped and people are being ridiculous with how much this will change things, at the very least this is going to be a huge new tool, and I think you're setting yourself up to be left behind if you aren't embracing this and learning how to leverage it.
Nvidia are the ones selling shovels in this gold rush, so it makes some sense that they'll make a lot of money out of it even if it was fool's gold all along.
Is Nvidias shovel-selling sustainable? Doubt it: when the gold rush is over, the demand for shovels will fall. However, we're long past the era when most money being pumped into the stockmarket was actually controlled by investors who cared about prospects beyond the next quarter, and it does make sense that speculative investors would be seeking to profit from the rise on Nvidias profits due to their shove-selling for this gold rush, even if later it falls back again.
Sure, but NFTs did not generate this many shovels being sold.
Which together with my point explains your own "But how do you go from NFTs, which never had widespread market support, to the market pumping a trillion dollars into Nvidia alone?" question.
Not only would LLMs and other more advanced generative AI have a significantly broader impact than NFTs if it lived up to the hype, but in technical terms it's much more dependent on GPUs for its functionality with any decent speed than NFTs as you can see in this comparison I just found with DDG.
Mind you, if you meant Bitcoin rather that NFTs (since the last big demand for GPUs was for Bitcoin mining rather than NFTs) the point that the possible impact of generative AI is much broader still explains it, plus if I remember it correctly Nvidia stock did got pulled up by the whole Bitcoin mining demand for GPUs (I vaguelly remember something about their share price doubling within a few years, but am not sure anymore).
Also keep in mind that Stock Markets at their most speculative end - i.e. Tech - have a huge herding effect: everybody wants to jump into the "next big thing" hype train as soon as possible and keep wanting to do so as long as it seems to be going (i.e. as long as they think there's a "Greater Fool" they can dump their overvalued stock on if an when it stops going) so there's a huge snowballing effect that pushes stock prices and company valuations far beyond anything explainable by actual and likely future financial improvements of their situation: this is how we get Tesla reaching a market valuation which is more that the rest of the Auto-Industry put together even though the former sells far fewer vehicles than the latter.
Stock Market rational considerations, especially in the most speculative parts of it, are not on "how much wealth can this company produce" but on "for how much more money can I sell this stock later", which is about one's own "smarts" or advantages that others don't have such as insider info and the gullibility of others, not about actual financial and accounting reasons of the company itself, which is why hype works so well to pump up the valuations of Tech companies.
The AI technology we're using isn't "new" the core idea is several decades old, only minor updates since then. We're just using more parallel processing and bigger datasets to brute force the "advances". So, no, it's not actually that new.
We need a big breakthrough in the technology for it to actually get anywhere. Without the breakthrough, we're going to burst the bubble once the hype dies down.
The landmark paper that ushered in the current boom in generative AI (Attention is all you need (Vaswani et al 2017)) is less than a decade old (and attention itself as a mechanism is from 2014), so I'm not sure where you are getting the idea that the core idea is "decades" old. Unless you are taking the core idea to mean neural networks, or digital computing?
I just don't get this. There has not been some huge leap in processing power over the past few years, but there has been in generative AI. parallel processing, on the other hand, has been around for decades.
I just don't know how one can look at this and think there hasn't been some big step forward in ai, but instead claim it's all processing power. I think it's pretty obvious that there has been some huge leap in the generative AI world.
Also I've been incorporating it more and more. It boggles my mind that someone would look at this and seea passing fad.