No announcement yet.

AI learning to recognise images without supervision

  • Filter
  • Time
  • Show
Clear All
new posts

    AI learning to recognise images without supervision

    we're just going to keep saying it can't happen until it's too late to have any control over the direction of this evolution
    The photos were used to help a Facebook algorithm learn to recognise images without supervision.

    Traditionally, algorithms have been trained on datasets which have already been categorised by humans - labelled cats, dogs or flowers, for example.

    But the Instagram photos were presented to the algorithm without the labelling.

    Afterwards it was able to correctly identify images with 84.5% accuracy, Facebook reported.

    Facebook has called its system Seer, an abbreviation of self-supervised.


      As long as it has an "off" button.


        Is it technically learning, though? As in gaining knowledge? What is it really? I don't understand. Has anyone elucidated upon what is actually happening here, or is it a mystery?


        • Vanilla Gorilla
          Vanilla Gorilla commented
          Editing a comment
          Mountains of code created by humans to give the illusion it's learning, mountains that one single human on the project doesn't see let alone understand

          It's not Artificial Intelligence, it's just a slighter cleverer way of pattern matching

        Wow. Thanks VG.

        That's impressive, but it's not life I guess.


          I was never claiming it was "life", just to be clear : )

          Whether AI ever matches "life" isn't so important, as whether it's indistinguishable to most people. Potential dangers don't rely on authentically replicating organic life.

          And at the same time, it's arrogant to not keep in mind we have no idea, really, what consciousness is. I think the most likely thing in the future is biological and technological merging.. at which point, if a human lends its consciousness to a machine that has superhuman abilities.. isn't it pretty much the same end result as if we created conscious robots? With the main difference being that hopefully the result would retain human priorities.


            It's not intelligent, its the wrong word. It's better to call it artificial calculating power, and albeit its impressive how it develops, it is not intelligent. It is not yet and likely won't any time soon be thinking for itself.
            The main thing we have to worry about in terms of control still lies with the programmer.