| Artificial Intelligence |
|
|
|
I first became acquainted with computer scientists in the mid-1960s, when I went to work in a computer systems lab. My assignment was to make computers more user-friendly. I soon found out that computer scientists, although a brilliant bunch, were able to communicate only with one other. They were already entranced with the idea of “artificial intelligence” and computer programs that “learn.” Now after about fifty years, the desire to develop such computer programs is just as strong, just as fruitless, and just as hopeless as ever, as indicated in “Let’s Make a Game” (July/August).
The article, for example, mentions the ringing of a doorbell. Whether we bother to answer it immediately will depend on who it is. For example, the ring of the mailman differs from that of the lady next door. During WWII I slept soundly through some bombings and shellings; however, I was immediately wide awake and on my feet taking cover in response to others’ waking me. On one occasion I was told that I had slept soundly through a six-hour barrage. The shock waves from this barrage were so strong they permanently deafened me to all sounds above 2500 hertz. The fact that I am here to write this letter attests to the correctness of my discriminations, whatever their basis. There is no area of study— from fine arts to theoretical physics or any form of finance, commerce, or industry—that has not profited from the use of computer assistance; however, let’s not call this computer assistance artificial intelligence or learning, because it is not. Furthermore, any attempt to duplicate behavior exactly, whether in a game or whatever, is doomed to failure because there is no way even to identify all the variables involved.
John Harry Hill ’49
|
||||