Announcement

Collapse
No announcement yet.

Do We Have An All-Purpose Artificial Intelligence Thread?

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

    Do We Have An All-Purpose Artificial Intelligence Thread?

    If not, this will have to do. This is a really good article written for the layman on the fundamentals of artificial intelligence and just what is the difference between machine learning and deep learning. There's some great stuff in there about the development of neural networks and just why it's so hard to develop what is termed a "strong" AI (i.e.: an artificial intelligence that can think and behave similar to humans).

    #2
    Tbh I think humanity's only hope is to build machines that can fix our problems and also treat us mercifully.

    Comment


      #3
      I wouldn't hold out much hope. We tend to introduce our own biases into the AI we create.

      Comment


        #4
        They might want amusing pets.

        Comment


          #5
          Machine learning, deep learning and AI have become the go-to bullshit terms for a lot of scientists who are simply running correlation analysis.

          Wake me up when a computer passes the mirror test.

          Comment


            #6
            Originally posted by anton pulisov View Post
            Machine learning, deep learning and AI have become the go-to bullshit terms for a lot of scientists who are simply running correlation analysis.
            How does that explain something like generative models?

            Comment


              #7
              As far as I know, those are algorithms that are fed a database of variables based on observations of past events, and based on probability calculatuons some kind of prediction can be made. A complex weather forecast model would be typical example.

              All of these models are ultimately built on the logic written by the programmers, and the database of past observations that is fed into them. So they are not acting autonomously.

              Of course, one can argue on a philosophical level what the hell intelligence and autonomy is. I mean, when I look at somebody's facial expressions I consult my own database of facial expressions built up over decades, and make a probability weighted estimate as to how that person is feeling, based on a logic script that was programmed into me by my family/society when I was growing up.

              Comment


                #8
                Lev Manovich's latest book on AI is available on Kindle for 1.99. Recommend.
                https://www.amazon.com/AI-Aesthetics.../dp/B07M8ZJMYG

                Comment


                  #9
                  https://twitter.com/chrislhayes/status/1117775101225832448

                  Comment


                    #10
                    Originally posted by Patrick Thistle View Post
                    They might want amusing pets.
                    Sony's new robot dog is out.

                    Comment


                      #11
                      Originally posted by anton pulisov View Post
                      As far as I know, those are algorithms that are fed a database of variables based on observations of past events, and based on probability calculatuons some kind of prediction can be made. A complex weather forecast model would be typical example.
                      That's accurate but it also underestimate their capabilities, to an extent. Generative models are fed data and using unsupervised learning can produce an entirely new dataset with variations - they're the technology that is used to generate specific kinds of computer music, for example. That goes far beyond simple "correlation analysis". One can argue that fundamentally that's what all artificial intelligence is built on but many human creative endeavours can be said to behave in a similar way. If someone asks me to design a table, I don't imagine what a table should look like based on absolutely nothing - it's not as if I've never seen a table before and don't know what one is. I look at examples of what I've been asked to build and I progress from there. Given my complete lack of any kind of DIY skills I would fail miserably. But if we imagine a machine that has the automotive capabilities to gather, assemble and manipulate raw materials and has been set the task of building a table then that machine has advantages that I don't have - impatience, tiredness and discouragement aren't factors, so it can repeat the process, learning from its mistakes, until it's made a table. You touch upon this with your other points.

                      Comment


                        #12
                        AI clearly has some way to go.

                        Comment

                        Working...
                        X