Hey there,
Bob's not my real name. To my right is a book called "Bobiverse". This may or may not be related to my username. I love sci-fi and just want A.G.I. to be here already (what's taking so long?). I'm 22 y/o, live in the Netherlands. I have finished my bachelor in Software Engineering and now I'm doing my pre-master in Computer Science.
I came to the conclusion that working on any side project would be a lot easier/more fun when you don't have to type everything yourself and could just ask. Even for just the "simple" stuff. So working on an A.I. before any other grand projects make sense to me. Not that I expect something in the near future. Anyway, it's quite fun to work on it.
I'm primarily interested in neuroscience and mimicking the brain. Or at least understand it to some degree and replicating the effects in a more resource-friendly version.
Everything I know about the brain is from my own lazy research. I have not (yet?) done anything neuroscience related at my university.
I think that we should treat A.I. as infants. The main benefit of having A.I. is that it's easier to hook them up to computer implants/interfaces (like what Neuralink tries to do with real people) because they... are computers. So that - and speed - is their major benefit compared to us. But like babies, we have to learn them everything step-by-step (and you don't give it a gun for f***'* sake).
I try to validate my models with what I know from my human experience and what I know of neuroscience. E.g. when laying in bed and everything is quiet I can pick up more and more small sounds. So it's plausible that there is some inhibition mechanism in place somewhere. There always seem to be holes in my knowledge so there are always assumptions.
Currently, I try to base my theory more on how smaller parts would function. And given that the smaller parts function in the way I hoped (which is testable), how would the bigger system turn out? I try to focus on a single part at a time but I try to keep it grounded by making assumptions which I can validate at some point. If that makes sense. This is a bit more practical when I actually have to program something.
I will post my theory sometime if it's not a huge success.
Hope to have some great conversations/theories/insights going on this forum.