Asklemmy

Screenshots should be against the Geneva Convention. Just post a link please and thank you.

You're absolutely right, I don't definitely don't think that we are there!

Although I do believe that humanity has always trended this way—starting with sitting on rocks, then shaping trees to fit the contours of our physical bodies as chairs. Now, we're trying to shape abstract knowledge and "thoughts" to fit the contours of our individual minds for similar reasons.

But with AI? I don't know, I don't see any stop sign ... Maybe that it never reaches this high mark we all expect?

I personally think that's the most likely outcome. Most of the advances lately rely on effectively "brute forcing" the problem space by shoving more training data in and by using more resources to calculate weights. There are minor improvements here and there by combining approaches, but development of new techniques has largely slowed to a crawl.

There's also still no clear path for any of this tech to make the massive leap from "trained for a purpose" to generalized knowledge, which is the most pointed to "selling point" for the whole idea.

And all of that is ignoring the fact that OpenAI, the biggest name in the space, operates at a considerable loss. They only still exist because Microsoft can afford to burn the equivalent of a small country's GDP on the small chance they get to be an industry leader on this. The resource, money, and energy investment for the current results are so absurdly mismatched that unless something huge manages to shake things up, I have a very hard time seeing it ever reach the heights the hype machine has been prophecizing.

Machine Learning is amazing, has been improving all sorts of things for multiple decades, and will continue to do so long after this current overhyped idea of AI fades away. The current glorified chat bots, generative AI stuff? I think we're already well past the point of reasonable ROI in terms of resources.

I'm not