Briefly, and apologies for the brevity...
I read this paper on a new programming language called Buzz that allows an abstracted control over robot swarms. Specifically, patterns of self-organized swarm behavior can be set up for groups of robots, and more directly, different types of robots within a group.
I'm interested. I want to try it out. I don't want to do with it real robots. I'd rather just set them up as black boxes on a computer simulation. More to the point, I'd like to create a specific kind of heterogenous bot swarm in the form of neural nets with these simulated bots and see what kids of patterns they'd create.
I doubt this will happen given my time constraints (at my dicking-around pace, I would if only I had another 150 years...), but it fits in nicely with the direction of thought I've been in lately.
A while back, Google caused a bit of stir with deep dreaming recursive visual imagery. The 'deep' part comes from a deep layered neural net that was used. Google, and many others, are reverse engineering brains with deep layered neural net programs. The nets are much more difficult to train than shallower nets of 3-4 layers of simulated neurons. Oftentimes, the neurons are simply treated at a higher level of abstraction as black boxes with generic inputs and outputs.
Hmm. Black boxes. Bots. Similar to what Buzz does, or claims to do. And neural nets, if they are to be truly reverse engineered, really need to take into account more than just neurons. The net is itself heterogenous, with different cell types, and - most important - pretty much supported by the glia.
Glial cells, it appears, seem to do more than just support neurons. They often seem to wield neurons. And the whole thing seems to be with swarming behaviors. So, brain like a swarm, not surprising, and worth pursuing.
More important, the brain alone is embodied, and the body is itself a very distinct heterogeneous swarm called a holobiont.
Holobionts have entered my news feed. Darwinian selection seems to occur not just at the level of the gene, or organism, but at the group holobiont level. This is unsavory news to plodding orthodox bishops of biology like Richard Dawkins.
Well, too bad. It's definitely where I think things are headed in biotech and AI. I really don't care about expert testimony to the contrary.