• nifty@lemmy.world
    link
    fedilink
    English
    arrow-up
    17
    arrow-down
    1
    ·
    5 months ago

    Balaji provides a great example of how not to think about developing systems that use AI. If you can build a system that is human-like in perception and cognition, then you don’t need anything more than what a human uses to navigate driving. His statement is like saying “I solved this problem because I didn’t actually attempt it”.

    • hglman@lemmy.ml
      link
      fedilink
      English
      arrow-up
      3
      arrow-down
      4
      ·
      5 months ago

      The problem stated is self driving cars, not use AI to self drive on roads unchanged from those used by human drivers.

      • nifty@lemmy.world
        link
        fedilink
        English
        arrow-up
        7
        arrow-down
        2
        ·
        edit-2
        5 months ago

        How is it self driving if it’s on fucking tracks my guy?

        Edit to say I get that “tracks” may not literally be required, but that’s besides the point

        Edit2 note that self driving as a term has been agreed to convey autonomous driving. If you’re thinking of trams or trains as self driving than you need to use the fucking correct terminology of self propelled. You cannot redefine a term because you failed the problem or don’t want to attempt it