Up@dawn 2.0

Wednesday, December 13, 2017

The Turing Test and the Chinese Room – 2nd Installment #10

            In my first installment, I focused on Alan Turing’s life and how much of an impact that the Turing Test has created over the last several decades.  To summarize the Turing Test: if a machine can fool a judge into thinking that it’s a human, it is “intelligent” by Turing’s logic.  However, is that necessarily the case?  Though computers have come a long way since Turing’s time, could a machine ever be as intelligent as a human?  This has been a subject of the debate in both computer science and philosophy ever since Turing shared his theory.
            In 1980, a philosopher from Berkeley, John Searle, argued that computers could never actually understand the questions that they were asked, but could easily answer them.  Searle’s argument, better known as the Chinese Room, has become just as famous as the Turing Test.  The Chinese Room thought experiment is as simple as the Turing Test: imagine you are in a room with nothing but a large assortment of manuals with matching symbols and a small opening in the wall.  Every so often, somebody would slide a piece of paper with a symbol on it through the opening in the wall.  You take the piece of paper, find the matching symbols in one of the manuals, copy it down, and slide it through the opening.  Now, imagine that the symbols aren’t just symbols, but in fact Chinese characters; to the people on the outside, you’ve been answering questions in Chinese, but you (probably) don’t understand anything that they’ve been asking you.  To Searle, this is exactly what a program trying to pass the Turing Test would do.  According to him, this proves that machines can never be intelligent.
            There are many critics to Searle’s Chinese Room argument.  Though most people agree that the Turing Test isn’t enough to show that a computer is intelligent, many believe that he is mistaken in thinking that this proves that machines can never show understanding.  For example, what if sensors (microphones, cameras) were added to a robotic body and it was allowed to learn by association, just like humans do?  This is referred to as a Total Turing Test, where if a robot could gain understanding through a similar environment as a human could, then it’s seen as intelligent.
            In the end, intelligence is a very difficult concept to be able to identify.  Humans are intelligent by nature, but labeling a man-made object such as a machine is not as simple as it seems.  Alan Turing first brought up the issue of artificial intelligence with the Turing Test, and with it brought an entire new conversation (or argument) into the world.  Years later, many critics emerged, including John Searle, whose Chinese Room argument is as well-known as the Turing Test itself.  I think that it won’t be easy (or possible) to satisfy everyone’s requirements for a high state of artificial intelligence, but it will come eventually.  I've included an article about two Facebook chatbots who were shut off after creating their own "language" that humans couldn't understand.  I've also included an open letter on A.I. signed by many experts on technology on the dangers of Artificial Intelligence.


Sources:

Comments made on other posts:

2 comments:

  1. Hello I am so delighted I located your blog, I really located you by mistake, while I was watching on google for something else, Anyways I am here now and could just like to say thank for a tremendous post and a all round entertaining website. Please do keep up the great work. chinese transcription

    ReplyDelete
  2. I truly like you're composing style, incredible data, thankyou for posting.
    Spanish audio transcription

    ReplyDelete

Note: Only a member of this blog may post a comment.