This sort of dovetails off a question I asked on the AI Exchange: Is There A Need For Stochastic Inputs To Mimic Real-World Biology And Environment?
I figured this site would give more insight on what I'm trying to accomplish neurologically/biologically.
My question is this: in order for artificial intelligence to more accurately simulate consciousness, bodily processes that may influence thoughts, and more complex response to stimuli that may not always be deterministic, would a new library similar to TensorFlow or PyTorch be necessary, or are these libraries sufficient enough?
This paper actually used FFNN to accomplish hunger/satiety: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4936290/
Would a new math be required, such as linear-time logic or fuzzy logic that would be better implemented in a completely new software library? Perhaps used in-tandem with something like TensorFlow, in order to reproduce the variations in feelings like hunger or random thoughts and creative inspiration?
For example, I'm thinking of an application similar to a tensor field, where over time a mood (happy/sad/angry) could be (literally) bit-by-bit changed, affecting surrounding tensors in a stochastic way, but it may not lead to a full mood swing at all.
Am I focused too specifically on a software library to accomplish the features of consciousness/moods/thoughts rather than the holistic Holy Grail everyone is chasing of accomplishing AI consciousness and artificial general intelligence? I was thinking of starting this as an open source software project but I may be on a wild goose chase.