🚀 go-pugleaf

Amsterdams AI Colloquium
Vrijdag 15 Januari
Zaal S 205,
Faculteit Wiskunde en Informatica,
Vrije Universiteit,
De Boelelaan 1081,
Amsterdam

!!!! Let op de ongebruikelijk dag:
!!!! Deze aflevering van het Amsterdams AI Colloquium
!!!! is op een vrijdag, en niet zoals gebruikelijk op woensdag.


	NORM-GOVERNED AGENT INTERACTION IN LAW AND COMPUTER SCIENCE
				    or
		 SHOULD COMPUTER SYSTEMS BE GIVEN RIGHTS?

			     Dr. Marek Sergot
			     Imperial College
				  London

For certain purposes -- and at the appropriate level of abstraction -- 
law, computer systems, and many other kinds of organisational structure 
may be viewed as instances of normative systems. The term refers to any 
collection of agents -- some or all of whom may be human, and some or 
all of which may be computer systems -- whose behaviour and interactions 
can usefully be regarded as governed by norms. Norms prescribe how 
agents ought to be behave, how they are permitted to behave, and what 
their rights are. Formal techniques developed originally in the analytic 
study of legal relations then find application in the characterisation 
of normative systems generally -- of which law is but one species.

This talk will expand on these general points, and illustrate how a 
specific set of formal tools from the analytic study of law may be 
applied to the analysis of some problems arising in the specification of 
computer systems -- concretely, the specification of access rights to 
sensitive medical information in a mental hospital. The formal tools are 
a modified version of the Kanger-Lindahl theory of normative positions. 
This is a formal theory which uses deontic logic -- the logic of 
obligation and permission -- and the logic of action to cast light on 
what legal philosophers have called the 'fundamental legal conceptions' 
of duty, right, privilege, and so on, leading to notions of 
responsibility and authority.

The final part of the talk will identify some particular issues that 
arise in the formal specification of computer systems. Of particular 
interest is the potential value of -- explicitly -- adopting the 
normative systems perspective on the specification of multi-agent 
computer systems. These last remarks will be illustrated by reference to 
the central role of 'commitment' in Shoham's Agent Oriented Programming.

The core of this talk is joint work with Andrew J.I. Jones, Department 
of Philosophy and NRCCL, University of Oslo.

This talk is aimed at both computer scientists and at Law-and-AI people.

Het Amsterdams AI Colloquium wordt georganiseerd door door de vakgroep
SWI van de UvA, de AI-groep van de VU, en is deze keer in samenwerking
met het Instituut voor Informatica en Recht van de Vrije Universiteit.

Voor meer informatie kunt u contact opnemen met
     Frank van Harmelen (frankh@swi.psy.uva.nl, 020-525 6791) of
     Frances Brazier    (frances@cs.vu.nl,      020-548 5588).
-- 
Frank van Harmelen			frankh@swi.psy.uva.nl
S.W.I.
University of Amsterdam
Roetersstraat 15,