Has the potential for this approach been stripped, or does it just have a different name now? There are matrices and feedback loops. What was the key insight that kept cybernetics from following machine learning towards neural nets?
Thanks in advance!
And while I’m plugging her books, her latest one is one of the best-written lay-audience books about machine learning and AI that I’ve seen. I strongly recommended it for anybody curious about the field, and also for people working in the area- her explanations are crystal-clear but not dumbed down in the slightest. It’s called “Artificial Intelligence: A Guide for the Thinking Person.”
The have a great podcast called Complexity: https://www.santafe.edu/culture/podcast
Link to description of that chapter: https://en.wikipedia.org/wiki/Cybernetics:_Or_Control_and_Co...
(I think it is the Dennis Gabor part)
Second-order cybernetics defines not only today's technological substrate (especially obvious in cybersecurity and risk modeling) but also contemporary culture.
I once gave a friend of mine Medina's "Cybernetic Revolutionaries", and I think he found it hard to get through because of the latter kind of cybernetics, even though he was very familiar with the former kind.
But see, that "other" part is also really a generalized/philosophical/speculative etc. take on control theory. "Systems theory" in the cybernetics-related sense is basically a name for 'this control theory stuff is so cool/impressive, let's see if we can generalize it beyond its standard engineering domain and draw any uselful conclusions from it.' Of course this kind of interdisciplinarity would be seen as standard nowadays, you wouldn't need to call attention to it by trying to name a new field.