Into the Grey Zone

I've been studying biological neural networks (bNN) for about ten to fifteen years, and artificial neural networks (aNN) for the past four or five (aNN were something I first learned about in Kutner circa 2005[2], oblivious to their future utility in machine learning). It's been interesting to observe how fast the field of machine learning has progressed aNN theory and application. IMO it has basically caught up to neurobiology - not so much as a body of knowledge but in the elucidation of the key components that allow NN to acquire and retrieve information (the modulation of synaptic weights, and efficient functions to update these weights). Machines are now smarter than ever, and have the capacity for adaptive behavior. A triumphant feat, no doubt.

But within all that progress, we still have no fucking clue what consciousness is about. We don't have a roadmap to get there. We don't know where it resides in the brain (somewhere distributed among cortex presumably). And we don't have a solid strategy for ML attempts at AI consciousness. The best we can offer is: "just keep building these things more powerful and more parallel and maybe it will eventually pop out". Who knows, maybe it will. But it probably doesn't work like that. For all we know consciousness may be a dirty trick, or nothing at all. Sure, we can think about thinking; it's a handy tool developed by billions of eukaryotic cells working together, to keep themselves alive (or at least to proliferate their shared dna).

1. https://en.wikipedia.org/wiki/The_Astonishing_Hypothesis 2. https://mysite.science.uottawa.ca/rkulik/mat3378/mat3378-tex...