I quitted a job paying me 240k post-tax as a fresh grad to do Full-time independent research in AGI
Just to share and hoping to get some feedbacks anonymously. What we are doing is a Mathematical model of consciousness and psychology. We figured out an alternative ML paradigm than DL. It learns much faster and more efficient than backpropagation, and it can compress information in a way that all current models cannot do. Attention-like mechanism, CNN-like structure, and more would self-organize within the model naturally. It can explain all the psychological phenomena we can currently think of / look for in a mathematical way. And it fits well on GPU, while it can also run on distributed network and analog computing. It scales linearly by computation while time is nearly constant. All the results would not be published since we are looking for direct commercialization. We are discussing with some VCs.