Open thread: when the next AI winter comes, what do you think it will look like? I think even those of us more bullish on the current trends in AI tech can recognize a massive hype bubble that is absolutely going to burst within the next year or two;, and those of us most bearish on its harmful elements think that there's *some* utility in parts of it. So I'm curious, what do *you* think is going to happen? What sticks around for good, what gets vaporized and forgotten, what limps along?
@glyph I think that by 5 years ago we reached the point where the horizon of what we can reliably predict is barely a year.
But still, here’s a possible scenario that we could avoid : LLM improvement beyond 2023 level will turn out to be slow, dull, minimal. Wider spread use of them will be for cost-“saving” encrapification, like automated telephone systems that prevent you reaching a real human, but worse. We will put people out of jobs and it will take15 years to admit that the net results were negative not positive
