Intelligence", which is a much less misleading name for the field.
To develop and deploy Universal Intelligent System in this
decade will require considerable development beyond
classical mathematical logic.
See the following for more information:
I don't disagree and I don't think that mathematical logic is the (only?) way to develop human-like artificial intelligence. Mathematical logic is the branch of mathematics that preoccupies itself with the proof of statements in a formal language. The ability to prove formal statements is probably something that is very useful to have, for any intelligent entity, but whether this ability is useful to create such an intelligent entity is another matter.
But I think we're all so far away from having a clue what an "artificially intelligent" entity might look like, or what "machine intelligence" might be like, that any predication about what may be needed to achieve it and how to come about it, is futile. We'll never know because we'll all be long dead by the time anyone creates a computer that can think like a human. Let alone one that can think better than a human.
Discourse using empirical information inevitably involves some disagreement requiring robust inference.
See the following article:
Oh, sure, McCarthy is one of my science heroes. But it is because of him that reserach that firmly belonged to mathematics and computer science in the broad sense, particularly the research on automated theorem proving, the natural continuation of the work of Gödel, Church and Turing, was lumbered with the goal of somehow showing that it can compute intelligence.
I mean, why is the ability to prove sentences in a formal languages a necessary requirement for intelligence? As far as I can tell, most humans do fine without it and so do basically all non-human animals that have anything remotely recognisable as intelligence.
Why is the ability to play chess a prerequisite of intelligence, for that matter, or the ability to predict protein structure? All those are typical tasks of AI as a field of research, because they happen to be "things that one can do with computers", i.e. computable computations. But what do they have to do with intelligence? What the hell even is intelligence? Turing gave us a model of computing machines and if one assumes that intelligence is a computable function, then it makes sense to try and compute it with a machine, but isn't that assumption the first thing we should test, before jumping right in and doing all the things we'd do if we knew it for sure to be true? And if it was true, wouldn't we have seen some progress towards that goal, after 70 years of study?
I'm speaking in this as someone who entered AI research because of my interest in logic programming and automated theorem proving, and mathematical logic in general. I am not interested in creating "AIs" one bit and yet, here I am, studying for a PhD in what is traditionally an AI subject. That's just dumb. Logic should have stayed where it belongs, in mathematics.
But... McCarthy was Church's student, so it was natural for him to set down the foundations of the field to the seminal work of his advisor, and others like him. Kind of like he was so keen on chess as an AI task- because he liked to play chess.
At least McCarthy's personal preferences did help to build a foundation of AI on theory, like I say in my comment above, so, for a while at least, it could have been less fuzzy and throw-stuff-at-the-wall-to-see-what-sticks-y than (most of) it is currently.
Yeess, that's true. John McCarthy named "AI". And he wisely set the foundations of the field in the work of his thesis advisor, Alonzo Church, and the mathematicians of his generation, and the ones before, all the way back to Gödel.
That is to say, McCarthy set the foundations of AI in mathematical logic, the branch of mathematics that began in antiquity by Aristotle, flourished in the Middle Ages, the Rennaissance and the Enlightenment with the works of Arab and Christian scholars, reached its peak with Leibniz and was finally exalted with the work of Frege, Russel, Whitehead, Hilbert, Tarski, Skolem, Herbrand, Gödel himself, Church, Turing and others. The branch of mathematics that gave us digital computers, much like Jurgen Schmidhuber points out in his article. It is this work that is the only, well, logical, foundation of artificial intelligence. That suffices to explain why the earlier "cyerbnetics" branch of research fell through: because unlike McCarthy's AI, it was not grounded in solid theoretcial foundations that had already yielded such monumental results in mathematics and computer science.
And it is the foundations of AI in logic that AI research must return to, eventually, because it makes no sense to throw out two thousand years of progress, just because an intellectual midget by the name of Sir James Lighthill wrote a report. It would be even worse to throw out such richess of knowledge in return for the ability to put bunny noses and ears on selfies with "neural networks".