the writer Nina Illingworth, whose work has been a constant source of inspiration, posted this excellent analysis of the reality of the AI bubble on Mastodon (featuring a shout-out to the recent articles on the subject from Amy Castor and @[email protected]):
Naw, I figured it out; they absolutely don’t care if AI doesn’t work.
They really don’t. They’re pot-committed; these dudes aren’t tech pioneers, they’re money muppets playing the bubble game. They are invested in increasing the valuation of their investments and cashing out, it’s literally a massive scam. Reading a bunch of stuff by Amy Castor and David Gerard finally got me there in terms of understanding it’s not real and they don’t care. From there it was pretty easy to apply a historical analysis of the last 10 bubbles, who profited, at which point in the cycle, and where the real money was made.
The plan is more or less to foist AI on establishment actors who don’t know their ass from their elbow, causing investment valuations to soar, and then cash the fuck out before anyone really realizes it’s total gibberish and unlikely to get better at the rate and speed they were promised.
Particularly in the media, it’s all about adoption and cashing out, not actually replacing media. Nobody making decisions and investments here, particularly wants an informed populace, after all.
the linked mastodon thread also has a very interesting post from an AI skeptic who used to work at Microsoft and seems to have gotten laid off for their skepticism
Hilarious.
Only five years ago no one in the computer science industry would have taken a bet that AI would be able to explain why a joke was funny or perform creative tasks.
Today that’s become so normalized that people are calling things thought to be literally impossible a speculative bubble because advancement that surprised everyone in the industry initially and then again with the next model a year later hasn’t moved fast enough?
The industry is still learning how to even use the tech.
This is like TV being invented in 1927 and then people in 1930 saying that it’s a bubble because it hasn’t grown as fast as they expected it to.
Did OP consider the work going on at literally every single tech college’s VC groups in optoelectronic neural networks and how that’s going to impact decoupling AI training and operation from Moore’s Law? I’m guessing no.
Near-perfect analysis, eh? By someone who read and regurgitated analysis by a journalist who writes for a living and may just have an inherent bias towards evaluating information on the future prospects of a technology positioned to replace writers?
We haven’t even had a public release of multimodal models yet.
This is about as near perfect of an analysis as smearing paint on oneself and rolling down a canvas on a hill.
That’s the exact opposite of a bubble, then. A bubble is when the valuation of some thing grows much faster than the utility it provides.
Yea sure maybe we’re still in the early stages with this stuff. We have gotten quite a bit further from back when the funny neural network was seeing and generating dog noses everywhere.
The reason it’s a bubble is because hypemongers like yourself are treating this tech like a literal miracle and serial grifters shoehorning it into everything like it’s the new money. Who wants shoelaces when you can have AI shoelaces, the shoelaces with AI! Formerly known as the blockchain shoelaces.
There’s a difference between a technology being a bubble and companies trying to use buzzwords to market goods.
Yes, 90% of the companies trying to capitalize on AI are going to go bust within the decade. But that’s because 90% of all companies don’t last 10 years.
The underlying technology will continue to be advancing rapidly though, and in world changing ways.
In fact, one of the biggest reasons why most current AI companies are doomed is because the tech is going to be advancing so quickly that they are building themselves into obsolescence sitting atop such quickly changing foundations.
Nvidia, Microsoft, Google, OpenAI - these guys aren’t going anywhere and are going to be continuing to make bank.
People repackaging up their products with a thin veneer of specialization are the ones that are screwed.
But this is again no different from pretty much every trend in history. Would you consider social media to have been a bubble because there were companies that entirely depended on theming your MySpace page or built upon Facebook marketplace that went out of business after short lived success?
Tulip mania overvalued something that didn’t have much underlying value. Just like blockchain cryptocurrencies.
AI isn’t that. The other month our company produced in a few weeks for less than a thousand bucks a project that kept being put off as it would have taken around a year of work at tens of thousands of dollars additional investment. And the end product was significantly better than we’d have expected from the manual process. There’s an inherent value in the core product of today’s generative AI even if the bottom feeders that circle every trend are similarly present trying to catch up scraps from unsuspecting marks.
Just like blockchain, right? That killer app’s coming any day now!
Not really. I’ve been bearish on crypto as a massively distributed pyramid scheme for over a decade now.
There’s a huge difference between speculative money exchange in a modern tulip mania and technology that’s actively being used and integrated across industries at scale at an unprecedented rate.
holy christ shut the fuck up
deleted by creator
Teach me!
uhh did OP consider my hopes and dreams, powered by the happiness of literally every single American child? im guessing no. what a buffoon
Iirc it still couldn’t do that, if you create variants of jokes it patterns matches it to the OG of the joke and fails.
Euh what, various creative tasks have been done by AI for a while now. Deepdream is almost a decade old now, and before that where were all kinds of procedural generation tools etc etc. Which could do the same as now, create a very limited set of creative things out of previous data. Same as AI now. This chatgpt cannot create a truly unique new sentence for example (A thing any of us here could easily do).
What ?
Of course it can, it’s randomly generating sentences. It’s probably better than humans at that. If you want more randomness at the cost of text coherence just increase the temperature.
People tried this and it just generated the same chatgpt trite.
you mean like a Markov chain?
These models are Markov chains yes. But many things are Markov chains, I’m not sure that describing these as Markov chains helps gain understanding.
The way these models generate text is iterative. They do it word by word. Every time they need to generate a word they will randomly select one from their vocabulary. The trick to generating coherent text is that different words are more likely to happen depending on the previous words.
For example for the sentence “that is a huge grey” the word elephant is more likely than flamingo.
The temperature is the way you select your word. If it is low you will always select the most likely word. Increasing the temperature will make the random choice more random giving each word a more equal chance.
Seeing as these models function randomly there is nothing preventing them from producing unique text. After all, something like jsbHsbe d dhebsUd is unique but not very interesting.
I don’t get particularly excited for algorithms from 1972 that come included with emacs, alongside Tetris and a janky text adventure but that is indeed the algorithm you’re rather excitedly describing
snore I guess
The pattern matching problem can typically be sidestepped by using emoji representations.
LLMs are trained by prediction, so one of the practical shortcomings is that slight token variations largely get ignored when the surrounding tokens are too similar to training data.
If you change the tokens to less common symbolic representations like emojis, the same issues that trip up naive variations often won’t.
Also, chain-of-thought prompting that gets the model to repeat the nuances in the variation can go a long way towards overcoming similarities to a normal form of the query.
A good example of this problem are variations on the “wolf, goat, and cabbage” problem to get to the other side of the river.
When GPT-4 first came out, initially there were a bunch of comments about how it couldn’t solve variations of the problem. But restating the prompt using “🐺, 🐐, 🥬” in place of the words and asking it to repeat any associated adjectives when mentioning the nouns as it worked step by step through the problem would have it solve variations correctly every attempt.
As I said earlier, part of the apparent shortcomings in SotA LLMs is that we’re still learning how to maximally use them.
That sounds perfect to me dawg
You have the insider clout of a 15 year old with a search engine
my god
Have a browse through some threads on this instance before you talk about what the “computer science industry” was thinking 5 years ago as if this is a group of infants.
If you feel open to it, consider why people who obviously enjoy computing, and know a lot about it, don’t share your enthusiasm for a particular group of tech products. Find the factors that make these things different.
You might still disagree, you might change your mind. Whatever the fuck happens, you’ll write more compelling posts than whatever the fuck this is.
You might even provoke constructive, grown-up, discussions.
I must note that the poster in question earned the fastest ever ban from this instance, as their post was a perfect storm of greasy smarmy bullshit that felt gross to read, and judging by their post history that’s unfortunately just how they engage with information
Oh good. Their history was why I relented and wrote something. A typical king shit.
you fucking idiot