[ExI] kalman filters are still dumb

spike at rainier66.com spike at rainier66.com
Wed Sep 7 05:29:16 UTC 2022


 

 

Regarding our recent discussion in this forum regarding artificial
intelligence, I suggested that if a Kalman filter is set up to make
predictions based on figuring out which of 1000 or more metrics are the 50
most significant, then it seems like we should agree that the software is
intelligent: it is going thru and finding an optimal solution, then
considers the top 50 observable metrics to predict something such as drought
in Europe based on accumulated cyclonic energy and other factors.

 

Now, after that discussion, I think we are best off comparing that Kalman
filter to chess.  Computers play great chess, which is a game about
predicting your opponent's next move.  It doesn't require intelligence to
play chess, since chess is a specialized form of computation.

 

Likewise, we can think of the weather predicting Kalman filter as a
specialized form of computation, clearly a really cool one which results in
the ability to predict weather very accurately but still not AI.  I will
stop whining about moving goal posts in the Kalman filter case.

 

spike

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20220906/f374ef32/attachment.htm>


More information about the extropy-chat mailing list