DANTE@EDWARDS-2060.ARPA (Mike Dante) (08/07/88)
Date: Thu, 4 Aug 88 12:03 EDT From: Mike Dante <DANTE@EDWARDS-2060.ARPA> Subject: Liar's paradox, AI .vs. human error To: AIList@AI.AI.MIT.EDU In-Reply-To: Message from "AIList Moderator Nick Papadakis <AIList-REQUEST@AI.AI.MIT.EDU>" of Wed 3 Aug 88 17:36:59-PDT 1. Bruce Nevin suggests that the solution to the "liar's paradox" lies in the self reference to an incomplete utterance. How would that analysis apply to the following pair of sentences? The next sentence I write will be true. The previous sentence is false. 2. Back when it appeared that the shooting down of the airliner in the Gulf was the result of an "AI" system error, there were a series of digests using the incident as a proof of the dangers of relying on computers to make decisions. Now that the latest analysis seems to show that the computer correctly identified the airliner but that the human operators erroneously interpreted the results, I look forward to an equally extensive series of postings pointing out that we should not leave such decisions to fallible humans but should rely on the AI systems. --- Or were the previous postings based more on presuppositions and political considerations than on an analysis of evidence? -------