Hubert Dreyfus on Decision-Making (1978)
ฝัง
- เผยแพร่เมื่อ 11 ก.พ. 2024
- Hubert Dreyfus discusses expertise and some of the attempts to make decision making more scientific in a talk given in 1978 at Iowa State University.
#philosophy #epistemology #artificialintelligence
Great channel to hear in my daily commute! Houston traffic not irritating me as much!
Thanks, I enjoyed your presentation! Formal here is used as an existential challenge to the Platonic form. Aristotle recognized the abstract nature of this and brought things down to earth in his Metaphysics. Both present a framework and identify the soul as essential to performing the task. Dreyfus's existential presentation dismisses logic, reason, and soul ( conscious ) to deal with it unconsciously in a free-for-all manner. An athlete' relies on muscle memory. Though they first learn the skill, and when performing it, they intuitively react through muscle memory rather than doing something different from the skills learned in the drills. Existential thinking works for the free thinker, though it has no place in the medical domain. The surgeon works relative to the learned function of the body, not just through his in-the-moment perspective. It's not a mescaline trip. It is the real thing .
Jung, when teaching dream analysis, would say to learn all you can about myths, archetypes and dreams, then forget about it all when analyzing a dream. So, not to impose a cookie-cutter solution and let the patient's dreams unconsciously speak for themselves. That, though is a mental process, a dream world that has yet no conscious reality and a broken stream of images. The conscious world does not work like that as it as it is ego dependant to keep the ship afloat.
I love Bert Dreyfus
He's very clear in this lecture and chess is a well chosen example
I think it's fascinating that his criticisms of chess programs of the era were spot on, and its greatly to his credit that he identifies the weaknesses in terms of "lacking real understanding" that would go on to hamper programs like (earlier versions of) Stockfish; its also to his credit that he identifies what it would take to create a program that does understand long range aspects of the game (i.e., self play over many examples)
I regret so much that he is dead. I wish i could hear what he thinks of new AI systems that do understand, that are not programmed with rigid rules but instead dominate the game through powerful intuitive evaluation trained by self play - eg, alphaZero, LelaZero
I think Bert was ultimately limited in failing to anticipate deep learning and machine learning in general, that he too narrowly interprets what "computer programs" can do (as he assumes they must be programmed) - but in putting the issue so sharply, he helps us see why this time around it really is different
it is true that expertise act on a sort of autopilot, but it is not so simple, reformulating the basics for yourself is useful, but you find that the real basics are not what it says in the books, or the manuals, the explanations given by experts or hobbyists, it is usually some rather different account of the basic principles, less principled and more detailed. for example you can take some simple motion in a sport, where if you go onto the internet you can find a lot of explanations of how to do basic things, and they are mostly a caricature of the proper motions, once you really master them, there are a lot of hidden detail, gauges of the motion as you do them, that is learned without noticing them explicitly in most cases, but they can be described, it is just that the real basics are often so much more complicated, that to turn the expert feel for the motion and proper adjustments into words and then back into imagination of the novice learner in terms of movement is complicated, and further complicated by the lack of awareness of their own coordination.
I think that Aristotle's distinction between episteme and phronesis is relevant here. All communicable knowledge has to be generalised, as speech is only capable of conveying general ideas, but experience is particular, which means that some amount of experience cannot be communicated.
Hi @philosophyoverdose may i ask you a question? When Dreyfus @ ~10:00 talks about "we have to check", does that make the issue conceptually related to the P=NP Millennium Prize?
I'm having hard time understanding the first 10 seconds.
Transcript helps👍
A “scientific” attitude about analyzing the world makes us distrustful of “intuition.” That is, folks that like to be “objective” about their observations of the world sometimes don’t leave any room for “unjustified intuition” in their observations because that would be too subjective.
Yeah but should I break up with my girlfriend or not?