Artificial Intelligence
Many decades ago, I borrowed a book from Sydney University’s library written by Cryptanalyst Alan Turing. Mr turing worked on deciphering crypted (coded) messages that Germany would send its front line Army during the second world war at the time of Europe occupation. Much later, after the war, he devised a theoretical experiment called The Turing Test . A theoretical experiment designed to test if an interviewer was able to distinguish if the answers he received from his questions were from a live person or a machine.
In 1996, IBM’s project Deep Blue would match its Algorithms with that of the World’s greatest Chess champion of the Day, a Russian called Kasparov. Deep Blue lost the first time. Having hacked Kasparov’s moves and having had its algorithms tweaked over the course of 6 matches, guess who won. Deep Blue did with 3.5 out of 6 matches (1 draw).
That was over 15 years ago. What happened since then?
In February 2011, you can find out if you watch Jeopardy. “Humanity’s best champions at the game show “Jeopardy”–Ken Jennings and Brad Rutter–are going up against the IBM supercomputer, named Watson, in the first of a three-night televised match.”
This time IBM’s machine will have to account for Context and Content. Implied meaning from which the words in themselves are meaningless unless the relative meaning of those words are accounted for within the sentence itself and from there the series of sentences that make up the questions being asked. Then there is the tempo and intonation, volume and pitch to name a few variables here … of the way people in normal conversation asks questions. During Jeopardy, Watson’s competitors will also be offering clues as to their level of confidence and hence this will tend to influence the “bets” they take before answering Jeopardy questions.
Whilst Watson will keep its “poker face” and not give away its next move, Ken and Brad may reveal their next move from the way they increase their bets. This live experiment from my perspective is not alarm as yet to prepare for the rise of machines but an opportunity to learn about ourselves to start with. In months to come, I intend to publish my first paper on The Binary Drivers of Human behaviour.
Thinking about Watson’s programming, I would hope that it would need more than just its vast computing power or mathematical logics to begin to emulate human behaviour. At the core of its programming, it would need to understand the key drivers of human behaviour itself. This subject matter takes me back many years to a time when I was consumed by Neuro Linguistics.
Whilst we cannot reduce Human conditioning down to its finite empirical drivers, there are indeed a few elements that possibly should be included in Watson’s programming such that it can understand some of the subtleties of human communication and its nuances and implied meaning. Exciting times as the hardware is there, the programming is more sophisticated than ever but is that all it takes to make a machine mimic a human to the point that it becomes totally indistinguishable for you and me?
