from Hacker News

Ask HN: Is the science depicted in A Thousand Brains, credible?

by rcshubhadeep on 8/22/22, 7:22 AM with 4 comments

I have been reading it for the last few days, although it is a very interesting read and as a ML fellow I could agree with some of the points, but I start to wonder to what degree is this research / science credible? Or is it another alternative approach that mainstream neuroscience community do not really consider (Something like The Physics Project by Wolfarm https://www.wolframphysics.org/)

I am asking this as mighty claims has been made and I do not know where can I see something that shows evidence about it. Also, if these claims are true then how come DL community is not more enthusiastic about them? I am really confused actually.

  • by mikewarot on 8/22/22, 8:49 AM

    It's a theory, if it leads to predictions that are proven in experiment, then it becomes part of science.

    My opinion is that it certainly seems plausible. Jeff Hawkins was recently interviewed by Lex Fridman[1], who has some background in AI research.

      1 - https://www.youtube.com/watch?v=Z1KwkpTUbkg

  • by hrorq on 8/22/22, 9:11 AM

    My general suggestion is to get a little more context about the understanding the brain to extract frameworks of intelligence.

    Two books general, popular books on the subject I recommend are:

    1/ The Master Algorithm - Pedro Domingos 2/ On Intelligence - Jeff Hawkins

    It's a good idea to read Jeff's previous book to see how he's built upon the original vision in A Thousand Brains.

  • by gardenfelder on 8/22/22, 5:42 PM

    My somewhat naive view is that Jeff's work is not like DL or much else in the ML field; it's like a fork in the road