AI (but not GOFAI) enthusiast here. Or rather "machine intelligence" as Numenta calls it. With a sciene/math educational background, I'm moonlighting as a natural philosopher, currently working my way through the "Learning the 3D Structure of the World" paper. I have a day job as a code monkey (yes I know, I'm sorry I'm young and need the money).
I'm also stupid beyond belief. Words can't describe my inability to comprehend the universe. Therefore my motivation to study and devleop AI is for AI to serve as a tool to augment the human intellect, not more and not less. I watched a video by Numenta where the stated goal was to create a "math savant" that solves scientific problems. This reflects the popular view of "an AI" being some sort of black box that acts in mysterious but (hopefully) ultimately beneficial ways. I find this idea very alienating and won't have any of it.
For that reason, in lockstep with developing AI, I am interested in developing the human-machine interface that makes handling that AI circuitry feel as if it was a part of oneself, to form a "human-AI hybrid". This doesn't have to mean sci-fi ways of plugging cables into one's head, it could be a well-conceptualized UI based on the I/O devices we have today. People and their smartphones today are already, while not quite human-AI, yet human-machine hybrids if you ask me.
I am also interested in how AIs or rather above mentioned human-AI hybrids could form networks, both in the technical and organizational sense of "network". I have a tangential interest in multi-agent systems and game theory. AI potentially blurs the line between humans and machines, upsetting some deeply held notions about the nature of the world and therefore being the subject of much popular fascination. However, improved networking abilities and the ability for multiple actors to effortlessly act as a single entity could force us to re-evaluate our view of the world at least as much and warrants investigation from early on.
"Like" this post if you think the same way.