Daniel Lemire's blog

, 4 min read

21 open problems in Artificial Intelligence

5 thoughts on “21 open problems in Artificial Intelligence”

  1. Peter Turney says:

    If you’re interested in self-reference and consciousness, then you will really enjoy Hofstadter’s latest book, “I Am A Strange Loop”, which is all about self-reference and consciousness.

  2. Mentifex says:

    AI has been solved.

    Quid est infra cum numeris Romanis?!

  3. vickey says:

    Problem 1: Representation in Design
    A fundamental problem for both artificial intelligence and design remains the one of
    representation. What is it that a designer knows and how do we get a computer to know
    it? Even if we are less concerned with what a human designer knows we are still left with
    the question of what needs to be known to design and how to get a computer to know it
    and use it.

    The early work on representing design knowledge as rules burgeoned into frames and
    semantic nets. More recently approaches based on conceptual schemas, conceptual
    graphs and distributed representations have been attempted. Whilst these approaches all
    add to our ability to represent, there is still a wide gap between what a designer ‘knows’
    when designing and what a computer-based design aid ‘knows’.

    Follow Below Location To Reffer More>>


  4. TeknoRapture says:

    Deep blue didn’t truly “learn chess”, in fact it cheated. For every play it generated several possible moves which a team of chess experts then chose from. It was more like a team of chess experts playing the champion, nothing more.

  5. DeepMind’s AlphaZero plays some incredible chess and routinely beats top engines like Stockfish. Nobody could beat AI in chess now.