Apology & further comments.
I apologize to all dyslexics who post here.
I also apologize to those for whom English is a second language. When I was very young, my father asked me to speak some Norwegian after I made fun of an immigrant. He then pointed out that the immigrant spoke two languages, while I only spoke one.
I do not apologize for suggesting that many Posters to this forum express naive opinions about esoteric subjects which they do not understand.
"Computers have been made using this sort of technology that, as a side effect of doing the tasks that they were built for, they were observed to dream while not in use. If this isn't coming close to conciousness then I don't know what is."
The above statement is fascinating if true, which I doubt. Sounds like a Twilight Zone episode to me, not something real. Perhaps it was from one of Nimroy's Pseudo-science TV shows. What computer program behaved as described above? Can anyone specify who built said computer? What company was responsible for the programming? How did they determine that the computer was dreaming?
My opinions on AI are based on study of various attempts to implement AI, mostly applied to various games. I have studied fairly detailed descriptions of the structure of Big Blue's programming. As a start, try reading "Computer Gamesmanship" by David Levy (Simon & Schuster). None of the AI efforts I have read about are structured around a huge number of If-Then-Else statements. Many use Von Neumann MiniMax strategy, simulated neural networks, game trees, position evaluation functions, post analysis of heuristic play, and various other techniques. The If-then-Else Statements are almost incidental to the overall structure of most AI programs described in various articles. If-Then-Else statements can be used as the basis for playing Tic-Tac-Toe, but are impractical as the basic structure for more complex games.
The major obstacle to AI is lack of an understanding of how the human brain functions. If you cannot describe what you want to do, it is difficult to write a program to do it. For example, current Bridge playing programs cannot do better than mediocre humans, while a chess playing program beat the world champion. Apparently there is no known description of a good basic approach to expert bridge play, comparable to the good position-evaluation function used by chess playing programs.
Does anybody have an opinion on how to decide whether or not some non-human entity has consciousness?