Tuesday, December 26, 2017

distributed intelligence

Michael Crichton's novel Prey, which is about swarming nanotechnology gone horribly wrong, is a pretty good primer on the concept of distributed intelligence, i.e., the idea that a corporate entity (e.g., an ant or termite colony) is composed of individuals that are fairly stupid when taken one at a time, but that are quite intelligent when acting in concert, as a whole. In the Joe Rogan video below, we learn about a Japanese experiment that seemingly demonstrated the "intelligence" of a certain fungus. The results are a bit creepy.

While the above interview might seem like cause for nervousness, I don't think that the form of intelligence on display by the fungus was much different from what we've seen happening in the field of artificial intelligence. This fungal feat wasn't on the level of human-fluent natural-language processing (which we don't have yet), nor was it a demonstration of artistic creativity. The fungus, in the tentative-then-relentless way in which it explored the subway network, was doing the organic form of what we see in supercomputers like Watson or Alpha Go: calculating and eliminating possibilities. There's no risk that the fungus will suddenly gain sentience and take over the world: its genetically encoded "algorithms," so to speak, work only for relatively simple tasks that can be addressed with raw processing power.

At the same time, if we stick to a basic definition of intelligence as "problem-solving ability," then we ought to at least respect the fungus for its ability to solve certain types of problems... although there is the question of whether the fungus actually realized it had encountered a problem, and then realized that it had solved it.

Nature is wondrous. Now lay off the psilocybin.

1 comment:

Anonymous said...

I agree with your view on what they were seeing. I am always a bit irritated with the AI people who are always claiming more than they really have. The main issue as I see it is that AI is algorithm driven, whereas living things are definitely non-algorithmic. Living things follow rules, but they bear little resemblance to the rules of computing. I think a major part of the difference is that computing is totally deterministic and living things always have a certain amount or randomness, or sudden non-determined change. Based on my study of nervous systems I think it fundamentally is due to the ratchet-like behavior, where change is step-wise not continuous. That's enough for now, we are starting to get into things I wrote long blog post on.

Merry Christmas and Happy New Year.