This video explains why LLMs can never solve the problems they face. It is really technical, but I still think it is well-worth a watch by everyone, with high school math knowledge
I was watching this too and, while I agree, the fact that this coming from religious apologists who work in “intelligent design” is some sweet irony that I wish I could explain. Apologists are famous for outputting bullshit and having terrible premises to go with GIGO. Perhaps they see their own image in the LLMs.
Define “work”. Define “think”. Define “reason”.
LLMs work. They’re working right now for billions of people. Model collapse is overhyped nonsense. Presenting a theoretical limit as “ai is so over” is dumb. They’re useful now, they’ll get more much more useful long before we hit any theoretical limits. This is as dumb as creationists saying “but evolution is so improbable!”
I am not going to say you are wrong. I just want you to watch the video and I really hope you would understand the core problems behind LLMs and the costs associated with it. If you don’t want to, that’s fine too. Facts don’t care about feelings. That’s the beauty of science. I also don’t want to say LLMs useless to everyone. But, it certainly is for me.
