David Deutsch wrote an interesting essay back in 2012 (http://aeon.co/magazine/technology/david-deutsch-artificial-intelligence/). His books follow similar themes and this article is a useful condensation of his ideas – most notable of which is the idea that intelligence/creativity/knowledge/universal computation is fundamentally about EXPLANATION – not instruction, not arithmetic, not reinforcement learning, etc.
He decries the lack of progress in artificial intelligence as due to flawed premises of the entire enterprise. He makes the case that “he human brain has capabilities that are, in some respects, far superior to those of all other known objects in the cosmos.” He declares self-awareness a thing and that universal computation can do it. And he concludes that all the ingredients for artificial intelligence are encoded in DNA we just need the right idea to unlock that and use whatever idea in other substrates to create other general intelligent entities.
I’m grossly simplifying the article which is a gross simplification of his books already but that’s because most of the details are irrelevant. Deutsch rightly ridicules the current prevailing approaches to AI and their inevitable failure to ultimately deliver intelligence. I agree with a lot of his reasoning around why AI with current behaviorist / inductive instructions approaches is doomed. But I disagree with him on pretty much everything else because he himself has built his arguments on flimsy premises. He assumes, as almost all scientists and philosophers and people, that knowledge is something. Something that is embodied, something that exists. I agree in a very simply way that anything that is learned must be learned through trial and error by the entity learning it and that learning doesn’t happen through transcription. But it’s not because knowledge results.
Knowledge is a reductive term that explains nothing and doesn’t really even clearly represent anything. It’s not a concept that can even be explained categorically or through endless descriptions. It simply is a general concept that sort of can be used to refer to lots of things.
Intelligence is the same kind of concept. It refers to nothing in particular. Self awareness, same. Good and Evil, same. Consciousness. Free Will. All the same.
AI won’t be coming because Real Intelligence isn’t a thing.
Learning is a slightly less reified concept than the others because it sort of gets at the point. The point of all computation and any perceived awareness is merely connectivity in a networking/graph theory sense. Advanced behaviors and “creativity” etc are merely effects of a hyper connected network. Learning is CONNECTIONS. “Knowledge” is CONNECTIONS.
Deutsch is correct that no one will be programming an AI. If something we might call AI comes to exist it won’t be because we specifically designed it. I would argue that it already exists, always has. It’s highly flawed to think that humans are the only things capable of awareness and thinking. It simply doesn’t add up. But that’s an argument for another post.
Everything is connected. Extensions of connections continue to evolve as more things connect to more other things in more ways. It’s such a simple, boring concept that it doesn’t seem that it would “EXPLAIN” it all. It doesn’t. It won’t. Because explanations are not the stuff of existence – intelligence or otherwise.
Thinking is not a thing. Thoughts are not things. There are connections between neurons and cells and organs and computers and planets and trees and galaxies and numbers and words and pictures and colors. Where one thing ends and another begins is very not clear…. even with the “laws of physics” which Mr. Deutsch uses almost exclusively to justify everything. The laws aren’t really laws. Ironically.
What Are We? There is no answer because it’s always changing.