So with the latest research of Ai actually sleeping to learn better, I can't help it and think that we're all being fed a thesis that most likely won't happen. But that doesn't min that the general life space is going to change forever and it's activelly changing as we speak.
We have these frontier models taking billions and soon trillions of freedom dollars to be trained and be able to reverse engineer copyrighted software, being able to hack hyper secure institutions in scary swarms, making it impossible for any expert team to protect us from it, solving century old math problems in days, making full playable games from one prompt, making enjoyable music, making full production grade movies, pretty much everything we can think of is now possible in a matter of days thanks to these frontier models.
But they still lack consciousness.
They still are not "intelligent by definition"
They still don't have emotions but can generate emotions in some people by manipulating them.
We're in an era wher AI can be used to produce incredible stuff but it can still tell you to kill yourself.
My question now is:
If models keep getting better at quality answers, perfect analysis, deterministic behaviours and user experience in general, what more are we going to need in order to make good digital products?
how much more valuable our prompts are going to have vs having councils of LLMs discussing continuous improvements based on data and research?
I still think that for us in order to trully enjoy a product, it has to come from a human perspective, that not everything needs to be optimal, that there's beauty and usefulness in chaos. I might be dead wrong, but I really don't such much more advancements in terms of capabilities in the tech space, pretty much everything we want to do, we can now do it in a matter of days. But it's never been harder to decide what to build.