Reply #5 to other comments posted about my proposal; February 1998
You've given me some things to think about.
a) I believe it could successfully "sense" environmental factors, if it constitutes a success to notice a discrepancy between current inputs (news feeds, ongoing publications?) fed to it and data already on file, and deciding what to do with that discrepancy. It is also conceivable that client software could monitor the data to see when data effecting results of a particular query is changed in a way that changes the results.
b) It could conceivably compare various interpretations to its data and see which one the data favors. I don't know how it would come up with multiple interpretations on its own, but it could come up with one interpretation, and react to others.
c) It would be able to correct its data based on the environmental factors, but it may require a human's prioritizing the "authority" of inputs (to resolve differences) or asking for a human's opinion in order to resolve conflicts. The changes in logic (or query output) would happen automatically as a consequence of that.
d) Translating that to action would have to depend on the decisions of its human customers, right? Maybe the client software mentioned in (a) would be one tool for that.
(Thanks for the link on speech technology. Microsoft's work also includes material on parsing out grammar for use by a variety of software types; I've discussed this with them a (very) little bit. It is good there are so many others as well doing related work; hopefuly one will be adequate soon, for this type of project.)