- From: Pat Hayes <phayes@ihmc.us>
- Date: Thu, 17 Jun 2004 19:44:48 -0500
- To: Graham Klyne <GK@ninebynine.org>
- Cc: "John Black" <JohnBlack@deltek.com>, "Jim Hendler" <hendler@cs.umd.edu>, <public-sw-meaning@w3.org>
> Or: is it possible, reasonable and practical to apply "speech act >theory" to automata? OK, let me grasp this nettle. In one sense it probably isn't, indeed: the software agents aren't 'little people' to which all of human social conventions apply, and they won't know Grice's maxims. But, there is a strong tradition in agent theory to the effect that what makes a software agent a genuine AGENT, as opposed to a mere scrap of code, is precisely that it does act ON BEHALF of a human agent, and in so acting as it were inherits some aspects of a social being, if only in a kind of borrowed way. If I send out an agent which makes assertions/promises/agreements then I am responsible for the assertions/promises/agreements it makes, not it: but OK, it is still doing the asserting/promising/agreeing on my behalf. So I want to make sure that all this socially-defined stuff that it is doing is clear, and the rules which surround such actions are clearly defined, just as I would if I were doing this stuff myself. In fact, more so, since I know that these little agents are pretty hopelessly lost when asked to do anything they havn't been built to do. And in this sense, I think that the SWeb vision DOES involve social actions performed by software agents. I'd be much happier if these stupid little thingies at least had the possibility of knowing what chunks of RDF they were supposed to trust. Pat <snip> -- --------------------------------------------------------------------- IHMC (850)434 8903 or (650)494 3973 home 40 South Alcaniz St. (850)202 4416 office Pensacola (850)202 4440 fax FL 32501 (850)291 0667 cell phayes@ihmc.us http://www.ihmc.us/users/phayes
Received on Thursday, 17 June 2004 20:44:32 UTC