Our main aim is to formalize these ideas in a way that will convince mathematical game-theorists of the poverty of their paradigm and the superiority of the AI view. We have a paper at the AAAI Spring Symposium and drafts for papers aimed at Theory and Decision and Negotiation Journal.
A web-site built by an REU student (Kay Hashimoto/Harvard econ) is at http://www.cs.wustl.edu/~kh/demo and exhibits how dialogue might be generated when two negotiating agents are each solving high-dimensional (10-30 variable) non-linear optimization problems to determine their utilities (the optimization is a proxy for an AI planning problem, and which might be more palatable to non-AI audiences).
2
3
4
The work is reported in a paper submitted to the Intl. Conf. on AI and Law. Related papers will appear in Law, Computers and AI, and Logic, Law and Computation and have been submitted to Ratio Juris. The site builds on work by REU students Joe Altepeter, Dan Pinkard, Jessica Linsday, and Mark Foltz (MIT AI Lab). It is http://www.cs.wustl.edu/~room5 (although it is often in a state of flux). NSF funding supported postdoctoral visits of Gerard Vreeswijk and Bart Verheij on this project.
The following tables are from the Room 5 site.