We talked in class yesterday about Artificial Intelligence -- a branch of Computer Science that tries to create machines with human intelligence. Here is the clip we watched from 2001: A Space Odyssey:
Notice how human HAL is in this clip. He displays a passive-aggressive tendency, refusing to answer Dave's question for almost a full minute, as Dave gets increasingly impatient. When HAL finally answers, he uses a calm voice as if nothing is wrong. He tries to avoid conflict; when Dave asks him what the problem is, his response is one of my favorite lines in the movie: "I think you know what the problem is as well as I do." HAL's last line of this clip is actually dismissive: "Dave, this conversation can serve no purpose any more."
One of the interesting aspects of the movie is whether HAL is displaying true intelligence or if he's just doing what he was programmed to do. HAL's programmed mission is to discover evidence of life on Jupiter's moon Tycho. He believes the crew is jeopardizing the mission because they discuss disconnecting him due to an error he committed, so he decides to kill them. Is this evidence of intelligence, or just a logical outcome of his programming, based on the instructions he was given? Computers are very good at doing what they are told to do.
This is the quandary of Artificial Intelligence -- is it even possible to make computers intelligent the way humans are? Or will they only ever complete tasks that are "complex", acting as a mediator between humans and the world? If a computer manages to learn something, is it only because the true intelligence is in the algorithm that humans embedded into the computer? Can a computer adapt and learn things we haven't trained it for? Can it go beyond its programming?
A good way of pondering these questions is to consider the state of what computers can do now. Consider Google. We expect that we can enter any search terms we want on Google's site, and it will instantly find relevant web pages for us. This seems intelligent. Google is not just blindly returning the thousands of pages that match our query, it is also having to make choices about which pages are most relevant. Google is certainly doing something we would find difficult to do on our own, and it appears to be doing some high-level reasoning about which pages are best.
Watch this video about how Google's search engine works:
Now what do you think? Does seeing what makes Google work change your idea of whether it is intelligent? Is it just following a prescribed set of steps that it was told to do? Is the real intelligence in the people who thought up this brilliant idea? Yet, if we showed this technology to someone 20 or 30 years in the past, they would certainly be impressed by how far computer intelligence has come. If Google doesn't yet meet your definition of intelligence, perhaps we are getting closer to cracking one of Computer Science's biggest challenges. Maybe artificial intelligence isn't unobtainable.
As you pursue your own learning in this class, keep HAL and Google in mind. Would your learning be possible without computer technology? Would it be less efficient? At what point does computing become so powerful, that the distinction between mediation and intelligence gets blurred?