For that to be useful, AI would have to evolve out of the current stage of being a parrot who got raped by an infinite amount of monkeys.
LLMs only know what we tell them. And they need human input, or they break.
https://www.cl.cam.ac.uk/~is410/Papers/dementia_arxiv.pdf
Right. So when you have an infinite amount of monkeys writing science papers nonstop, injecting that data into a blockchain, AI can then use that for a brain. At that point it has a perfect memory that can't be manipulated by a king monkey that likes to control information and decide who hears what and how.
I know AI can't tell you today's news until humans publish it. It's expected to evolve but will never be able to create current events and discoveries. A social network like X has millions of people working for AI right now, volunteering their time to help create and train an AI brain, and they don't even know it. Some even pay X to do this work because they think check marks are shiny. Then the masters of X decide what their little robot can remember. Other networks and corporations are doing the same. It's a disaster in the making.
Talking to AI will replace search engines. That's what I meant when I said majority will go to AI for answers. So it would be beneficial if the answers are in a safe place.