Let’s imagine that AI will become cheaper and more efficient, it will not differ from humans in terms of the quality of its work, it will replace almost all intellectual workers, and only the operators of these AI models will have jobs, that is, one person or several people monitor the entire office of AI workers for a small salary. Yes, the AI bubble will burst, but the problems of ordinary people will only get worse from this, jobs will not return, no, automation will continue anyway.
Is it worth retraining as a mechanic, plumber or something like that?
Nah.
Most technologies become exponentially more difficult to improve. We’re climbing thar curve now with gen ai but its still so far from being usable i honestly doubt it will ever replace a significant portion of tasks comprising knowledge jobs beyond the most basic.
Like self driving cars. They were almost ready years ago but the last 5% of capability seems to still be out of reach.
Gsm AI doesn’t just hallucinate sometimes. In my “knowledge job” you can load a few thousand words of reference materials into context, ask questions about it, and anything that requires even a modicum of inferred meaning is very likely incorrect. Worse, bots are amazingly good at being xlconfidently incorrect. I dont just mean not quite right on a few technicalities. I mean the opposite of correct.
My understanding is that the current state of physical robots is much worse. We can barely get a bot to drive around your house sucking up pet dander. I watched some commentary recently that made a very convincing argument that humanoid robots will never be popular for any purpose. Just imagine a bot of any form assigned to some simple task like clean a bathroom, or weed a garden - its comical.
Just like flying cars were “the future” until it became obvious that its a terrible idea, I dont think bots will replace many physical jobs.
We’re barely past the Wright brothers’ plane stage of AI right now, so predicting when the technology will level off is very difficult and just extrapolating forward from the limitations that the technology has today is unlikely to be reliable. I think you’re right regarding what we’ll see in the next few years but I have about thirty years until retirement. By then things will probably be very different and that’s something to keep in mind when choosing a career (or deciding to have a child).
Well, you’re right that the future might contain unexpected developments, and extrapolating forward is not going to be “accurate”, but my comment describes the trajectory that we’ve seen in other technologies.
With the greatest respect, the reference to the wright brothers is a hyperbolic metaphor. I suspect that the wright brothers could see that they had achieved a breakthrough moment and they didn’t know what the limitations of this new technology might be at that time. With gen AI I think we have already encountered the steep curve of diminishing returns.
The improvements in the last several years have been modest and incremental, but the infrastructure required to achieve them has been gargantuan. My point being that we have already reached that point of diminishing returns.