I honestly think the biggest absurdity around AI is that it's currently the basis for the largest financial bubble in history, and yet (outside of a handful of very specific applications in medicine), it doesn't do much of anything. Investors, politicians, the 'adults in the room' were just this impressed by a text generator.
How did you first use ChatGPT? 90% of people asked it a stupid question like "How would The Graduate be a different movie if Dustin Hoffman's character was played by Grimace?" and then giggled when it gave you a serious response. Have you done anything more profound with it than that since? Has the novelty of that worn off for you yet?
Unless you have bad intentions (having it do an assignment for you) or are fundamentally misusing it in a dangerous way (talking to it instead of consulting a therapist, or using it as a stand-in for a real human relationship), what we have left is a product that's fun to dick around with every once in a while.
It's not the printing press, or the car, or even the Internet. It is not a society-transforming innovation if we limit the scope of its use to 'desirable' applications. We're inserting it into everything because we're trying to intellectually justify why we should care about this beyond the initial novelty of a program you can have a convincing facsimile of a conversation with. But we haven't. Our global cultural response to AI is completely and utterly out of proportion to what it actually does.
This isn't even the first time in the past five years the market has gone all-in on what is essentially a toy for adults. Remember NFTs? Remember how you completely stopped hearing about them the moment something shiny and new (AI) came along? This is indistinguishable from the behaviour of children. "I don't wanna play with you anymore..."
AI isn't making money because you ask it silly questions. It makes money by removing redundant task or increasing efficiency of a task. That's why we are integrating machine learning in to medical devices such as insulin pumps that automatically adjust insulin based on real time data.
I think your post indicates you aren't exposed to the actual use cases and think it's just for the general public to ask questions. That is 100% not the use and is just extra data retrevial.
I hate the tech sphere of money as much as anyone else here, but I think the issues with AI is the same issues with all idnsury. New technology and information concentrated in the hands of a few.
This is going to get hated around here, but LLMs are actually a lot more useful than just being silly toys.
I use it for work regularly and it produces reports comparable to what I would expect from a entry level engineer. It has problems that have to be fixed during review for sure, but so do entry level engineers.
Now that's not to say that it's ultimately going to prove it's multi-trillion dollar investment value. If it doesn't progress significantly, and very quickly, than these companies will start to run out of investment money.
But that's how the "free market" is supposed to work. Investors invest in something to develop the idea. If it doesn't work then the investors lose their money.
The thing that's broken in our economy is that the federal government isn't allowing the market to make corrections. For every correction since the dot-com bubble burst the federal government has swooped in and bailed out the bag investments, preventing the correction to occur. And so, for the past 15 years, investors have just moved forward with the assumption that if there is a correction then the federal government will bail them out, so there is no reason to ever pull back on investing because there is no risk anymore, which creates a self-fullfilling prophecy and prevents corrections.
It's sort of looks like we've accidentally figured out how to cure recessions.