I heard that there would be a new Great Depression.
I’ve also heard a lot of different theories, and one of them interested me: after the bursting of the bubble, AI will not disappear anywhere, is it in vain that so many data centers were built? AI will be embedded everywhere, people won’t even be able to test how it works, “it works somehow, great!” because all human workers will be fired, not at once, of course, but gradually, In a few years, about 2-7 years to be exact(depending on the industry). And because of this, AI systems will begin to get out of control, this will cause incorrect diagnoses, failures in the banking system, arresting or killing innocent people by drones, and so on.
The reason I’ve explained this so poorly is because, in the first place, the topic of AI and the debate around it is terribly infuriating to me, and it’s obvious that the harm from AI won’t be able to compensate for the little benefit it will bring. Secondly, I use a translator, so my text may seem crooked, unnatural, or silly.
It’s already happening. A lot of services and free tiers got cut in the last months, and paid tiers got more expensive.
It’ll not completely vanish, if you’re a sane human being you probably know it has a few usages here and there.
But once you start using it for real work you’re done, it’s just a matter of time for the chaos.
I think we’re in the stage of people slowly realizing that AI work isn’t worth the price compared to a real human work. Because it’s expensive for a not so well done job. While real worker do a solid job, with consistency, for the same or less price.
The irony, given that machines always did jobs faster and more consistenly (like a calculator). But this case here we’re talking about a machine that try to mimic humans work without a well defined behavior.
LLM models are evolving very little these past months, very few improvements every new release. And so far no solid proof that it can do things well as humans do.
That Anthropic attempt on making agents build a C compiler was catastrophic, an insane amount of money for a thing filled with flaws. But proved it can’t do anything true valuable and trustworthy.
This is a good summary.
One thing I will add though, I don’t think AI is going away any time soon. The hype and saturation will probably fade a bit (hopefully). But what we are also seeing right now is a move to more self-hosted AIs due to price increases in things like Copilot.
I wish some if my coworkers would realize the limitations and hazards of over-using AI. Too many of them are still deep-throating all things AI, pushing for it to be used for every single little thing.