[cjs_ac]

The bull case is that when the venture capitalists stop subsidising LLM providers and expect them to turn a profit, the actual end-user costs exceeds the cost of employing a human. I don't know whether this is actually true, but it might happen.

[empath75]

Even it AI is better and more cost efficient at doing everything that humans do, there will still be work for humans to do. AI development will focus on stuff that ai is best and most efficient at. There will be many things that AI may be better at than humans but nevertheless would not be the best use of AI, and humans can still do that work.

[drivebyhooting]

If AI is indeed better than humans in a particular task, why wouldn’t it be applied to that case? There is no fundamental limit to how widely AI can scale and be deployed - only capital restriction.

Sure higher ROI applications will be chased gist, but in time even the least of tasks would be subsumed by AI.

[Zigurd]

Either that, or the AI bubble bursts hard and those people lose their jobs too, and nobody has a prospect of getting their job back. That then causes the market to lose enough that it becomes impossible for PE firms to exit any of their investments.

[trevithick]

True, my question assumed AI "progress" and adoption follow the hype trajectory. Reality could be closer to the scenario you laid out. The bubble pops, some AI tools maybe improve things in some areas, societal disintegration gets kicked down the road a few years.