Intelligent Interactive Distributed Systems

1
Are Law-Abiding Agents Realistic?

• Frances Brazier
• Onno Kubbe
• Anja Oskamp
• Niek Wijngaards

2
Interdisciplinary Research

• Computer Science (CS)
• Artificial Intelligence (AI)
• Law

3
Research at Vrije Universiteit Amsterdam

• Two research groups
• Intelligent Interactive Distributed Systems (AI
  CS)
• Computer and Law
• Intensive cooperation for a number of years

4
Agent characteristics

• are autonomous and pro-active
• are often intelligent
• can learn
• have knowledge, can be experts in a domain
• can reason
• manage their own resources planning, scheduling,
• communicate with other agents
• interact with the objects and/or services
• and
• may be mobile (roam the Internet)

5
Internet

• Highly dynamic environment
• Large-scale system
• too large for centralistic administration

6
Examples of tasks agents can perform

• Search the internet and intranets for information
• Process the information
• Combine different types of knowledge e.g. build
  profiles
• Manage distributed knowledge e.g. version
  control
• Plan, schedule and monitor (distributed)
  processes
• Plan, schedule and monitor (distributed) projects
• Perform (n-party) transactions

7
New technology required

• Agentscape is a framework for
• scalable,
• interactive,
• secure,
• robust,
• distributed agent systems.

Currently being designed and implemented by
the IIDS Group together with (Tanenbaum and
van Steen (both Computer Systems Group ), VUA
8
AgentScape Framework

• Includes
• a distributed agent operating system (AOS)
• services
• agent management
• matchmaking
• directory services
• agent modification (e.g. agent factory, cloning).

9
New legal frame of reference required

• Are agents full-fledged citizens of todays
  society?
• What rights do they have?
• Obligations?
• What is their legal status?
• Are all transactions in which they are involved
  legal? Valid?

10
Legal Requirements

• Traceability
• Privacy
• Identity
• Right to perform transactions
• Legal status
• Trusted third parties

11
Related Models and Tools

• Trust Models
• agents, owners, machines and locations
• Authentication and authorisation
• Audit trails
• Life-cycle management
• Protocols
• System level
• reliability, fault-tolerance, resource guarantees

12
Example Trust and Cloning

• Locations have different levels of trust
• Agent refuses to go to untrusted locations

• Agent may
• clone itself
• send helper-agent
• Legal Issues of helper
• Identity?
• Legal Status?

Home
Bookstore
13
We need bridges!
Computer Science
Artificial Intelligence
Law
14
The ALIAS project

• Analysing Legal Implications of Agent Systems
• Interdisciplinary Research Project
• IIDS Group (prof Frances Brazier)
• Computer and Law Institute (prof Anja Oskamp)
• Center for Law, Public Administration and
  Informatization (prof Corien Prins)
• Aim
• Cookbook How to address legal issues for
  agent-developers

15
We need your input!

• Here
• Website and discussion forum
• http//www.iids.org/alias/
• http//soapbox.cs.vu.nl/ALIAS/
• Email
• Niek Wijngaards niek_at_cs.vu.nl