On Computers ‘Learning’

Briggs is certainly right that much of what is touted as ‘Artificial Intelligence’ is just the use of electronic machinery to implement the calculations in a method devised by human intelligence. Indeed, for now that really is all that computers can do. But it already goes far beyond playing optimally at tic-tac-toe (trivial) , or chess, or simple kinds of poker. Those are what attract popular attention, but they imply far less for the future than do even the earliest attempts at voice and image recognition./p>

A true “learning” program isn’t just the implementation of a previously worked out method of solution to a particular problem, but takes as its input the results of its various earlier responses to similar problems and from that constructs a better solution algorithm than the one it had before[*]. This can be done deterministically by a machine, and perhaps it is only hubris that convinces some of us that what we do is qualitatively different from that.

So I think Briggs’ commenter Larry Geiger nailed it with “The extrapolation from what computers can do to what some people *think* that they can do is foolishness”(emphasis added).

[*]Update(2015-01-13) – like this which coincidentally came to my attention on the morning after I wrote the above.

This entry was posted in uncategorized. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *