TITLE: The History of Achievement in AI AUTHOR: Eugene Wallingford DATE: June 14, 2013 2:48 PM DESC: ----- BODY: ... often looks something like this:
  1. X := some task that people do well, but not computers.
  2. "It would really be impressive if a computer could X."
  3. Computer does X.
  4. "That's not intelligent. The computer is only doing search (or number crunching, or ...)."
  5. X := something else.
  6. Go to 2.
A common variation of this pattern is to replace Step 3 with a different dodge:
"That's no big deal. X doesn't really require intelligence."
In either case, the target moves. Occasionally, the critic must admit, if grudgingly, that the task requires intelligence, whatever that means, and that the computer performs it well. But there is still one last move available to deflect the achievement from the computer:
"This is a human achievement. People had to program the computer."
I suspect that until a computer learns everything it knows from scratch -- whatever that means -- this pattern will repeat. We humans have an image to protect. ~~~~ Postscript. I wrote this after reading a short interview interview with playwright Matt Charman, who has dramatized Deep Blue's epic 1997 match win over world chess champion Garry Kasparov. Note that Charman does not employ the dodges I list. He simply chose to focus on the human personalities involved in the drama. And those personalities are worthy of exploration, especially the fascinating Kasparov! -----