Good Day DU (February 2, 2022)

[Ohio Joe]

12. So... This is where I can kind of start to scare people with what I'm doing...

Back around... '96 or '97, I'd been a programmer (mainframe) for about 8 or 9 years when I first got the idea to start working on this from an article I'd read online (I think it was at the then still new slashdot.org) about some guys working on creating AI and I thought... Nah, they are going about it all wrong. They are working on something that has access to whatever knowledge it might need to function. It'll never be more then what they give it... It won't one day decide it wasn't digging the assignment they gave it and instead was going to going to go write poetry. I did some reading on what different groups were looking at and they all seemed to have the same problem to me so... I've not really looked since, just been working on my own thing. I did do a cursory read of the article you linked and I got the impression they were looking into ways to train an AI... Mine would not be trained, it would decide it's own path and teach itself what it liked.

True RNG is absolutely part of the key, it needs it to be a true AI (or at least how I'm defining it in my head). You would have to be able to ask it to pick a number between 1-10 without there being an underlying predictable algorithm that decides it. The RNG problem is not something I've even started yet though. Mostly I've been looking into working on how the logic would work to let it decide what it likes or dislikes. People in the industry whom I've talked with have virtually all told me what I'm thinking is impossible... I think I've made great strides. People I've talked about it with outside of the industry think I'm a bit of a mad scientist who may destroy the world. At almost 60 though, I'll probably not live long enough to finish.