The AI Bubble

[beancounter]

According to this person, AI isn't delivering the it's promises, and is thinking of resorting to ads and porn to sustain itself. Apparently it's a "circular" industry where all of the inside players are just passing money between themselves giving the illusion of a revenue stream with little external revenue sources.

So, in short, AI will not take over the world. It's a bubble.

 

[Second Summer]

Absolutely, it's a bubble, and I just can't wait for it to burst.

And the problem isn't infrastructure, of course, except in the sense that we have wasted much too much money on it. The problem is the, so far, lack of viable business ideas, or just ideas for how it can be used to improve people's lives, or how it can be used in non-revenue-generating sciences such as astronomy etc.

We suddenly have this truly amazing technology, but we don't know how to make use of it. We have the answer but not the question.

I've been faced with a similar problem at my job: All of a sudden, our database systems can now integrate with AI (LLMs, specifically), but what is the actual use-case going to be? That's much harder to answer than how to use these systems, find the data centres etc. And especially when people are already just happy with the systems/applications that we have, so they don't want to devote time and effort on it.

However, I think over time we will start finding use-cases for AI. It's just going to take much longer and adoption will be slower.

So yes, another tech hype bites the dust. To a large extent, at least.

 

[Lou]

And the problem isn't infrastructure, of course, except in the sense that we have wasted much too much money on it. The problem is the, so far, lack of viable business ideas,

What lack. there are more examples than I care to list here.

or just ideas for how it can be used to improve people's lives,

Siri, Alexa, navigation, medical diagnosis.
again. more examples than I care to list.

or how it can be used in non-revenue-generating sciences such as astronomy etc.

Since you mentioned astronomy. How about Data processing the huge amount of data the big telescopes generate. or image enhancement.

 

[Second Summer]

What lack. there are more examples than I care to list here.

I must not have been very clear, sorry: Yes, there are many good (and bad) use-cases. But it's far from enough to justify the enormous hype and resources diverted to it.

 

[LettuceBe]

It's an investment bubble but the tech will still exist when/if it bursts. Just like when the dot com bubble burst. Loads of investment money was lost but by then the infrastructure and tech had been developed that still benefit us today.

I think we all focus on the chat apps when we think of AI but there's a huge side that most of us don't see. Medical applications are a massive one that could be worth an enormous amount of money.

 

[Lou]

Our history is littered with examples of technologies that people predicted would fail or could not be monetized.
Remember YouTube. No one thought it was a good investment. (except for Google who had money to burn)
How bout Apple when Jobs left.
And GMs EV 1
MapQuest.

 

[Damo]

My machine has a fairly high-end gaming graphics card, while it can only run small models that fit on my puny 10GB of video memory, it uses a lot of power when regurgitating an output, hallucinating almost every time, regardless of which model I use.

I hate to think of the amount of power it's taking to generate responses you see with every Google search, every ChatGPT prompt, it cannot be sustainable...

 

[PTree15]

My friend, who is a researcher for a magazine, is worried that AI will take her job. Her company just announced that Chat GPT is coming to her office, and she believes it might learn enough to put her and a number of colleagues out of jobs. She figures it will learn from the workers and eventually learn enough to replace them.

 

[LettuceBe]

I think we all focus on the chat apps when we think of AI but there's a huge side that most of us don't see. Medical applications are a massive one that could be worth an enormous amount of money.

I've changed my mind somewhat about this. From what I've been reading, the vast majority of investment is in LLM's (chat bots).

This is unfathomably stupid. There is no way all that investment can ever be recouped. I don't see it as a step towards AGI at all. There'll be enormous computing power by the time the bubble bursts. I just hope someone comes up with something better to use it for.