Your premise is wrong. You assume these people are spending money creating useful output or would otherwise understand and be able to implement a more efficient means on their own. In the words of David Graeber a lot of people have bullshit jobs, are you sure the "AI" isn't alleviating some other problem for them?
> and not see the possibilities it opens up.
The current technology has no natural exponential growth curve. Which means for a linear increase in spending you get a linear increase in accuracy. Any thinking person should see where this is going. Which is why you should call these LLMs so you don't accidentally fool yourself.
I mean, of course, when AGI does arrive and has a reasonable power budget, then we're talking. The current technology will never become this or anything like this. This will almost certainly lead to a new "AI winter" before AGI happens and will likewise almost certainly not occur during yours or my lifetime.
If you do believe that then I have a self driving battery powered semi to sell you that's fully autonomous and will run road cargo trains for you all day and night for huge profits.