Chapter 25: Challenges and Extensions

1. Chapter 25:Challenges and Extensions Service-Oriented Computing: Semantics, Processes, Agents– Munindar P. Singh and Michael N. Huhns, Wiley, 2005

2. Highlights of this Chapter • Trust • Ethics • Coherence • Benevolence • Managing Privacy • Key Challenges and Recommendations Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

3. Systemic Trust • Fundamentally • The information agents retrieve must be accurate, or characterized accurately • The information agents contribute must be used appropriately • Requires • Sources have reliability and reputation, and specify constraints on usage • Dependencies are preserved and maintained • Results: information items have credibility and domains of utility; agents self-organize into service communities Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

4. Trust Trust means that services • Are understood in context • Have the right capabilities and understanding of needs • Follow legal contracts where specified • Support one’s organization or society • Follow an understood ethics • Failing all else, behave rationally Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

5. Ethical Abstractions Agents that are members of a society must have an ethics and a philosophy. This requires the development of components for • Deontological ethics • Teleological ethics • Consequentialism • Duties • Obligations • Applying ethics Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

6. Motivation The ethical abstractions help us specify agents who would act appropriately • Intuitively, ethics is just the basic way of distinguishing right from wrong • It is difficult to separate ethics entirely from legal, social, or even economic considerations Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

7. Right and Good • Right: that which is right in itself • Good: that which is good or valuable for someone or for some end Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

8. Deontological vs. Teleological • Deontological theories • Right before good • Being good does not mean being right • Ends do not justify means • Teleological theories • Good before right • Something is right only if it maximizes the good • Ends justify means Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

9. Deontological Theories • Constraints • Negatively formulated • Narrowly framed • E.g., lying is not not-telling-the-truth • Narrowly directed at the agent’s specific action • Not its occurrence by other means • Not the consequences that are not explicitly chosen, i.e., only applies on consequences that are explicitly identified Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

10. Double Effect Deontological theories distinguish intentional effects from foreseen consequences • An action is not wrong unless the agent explicitly intends for it to do wrong • Legitimizes inaction even when inaction has predictable (but unintended) effects • Shut down bank ATM for diagnostics even if that might leave someone without cash Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

11. Kant’s Categorical Imperative • Whatever maxim an agent uses must be universalizable, i.e., in the society of agents • Respect for others (no lying or coercion) so they can consent • False promising is unacceptable, because if everyone did so, society would not function • Agents’ maxims are uncertainly inferred from their actions Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

12. Teleological Theories • Based on how actions satisfy various goals, not their intrinsic rightness • Comparison-based • Preference-based Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

13. Consequentialism An agent should promote whatever values it adopts • Actions are instrumental in the promotion • Honor the values only if doing so promotes them Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

14. Utilitarianism This is the view that a moral action is one that is useful • Must be good for someone • Good may be interpreted as • Pleasure: hedonism • Preference satisfaction: microeconomic rationalism (assumes each agent knows its preferences) • Interest satisfaction: welfare utilitarianism • Aesthetic ideals: ideal utilitarianism Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

15. Prima Facie Duties • What agents need to decide actions are • Not just universal principles (each can be stretched) • Not just consequences • But also a regard for their promises and duties • Agents have prima facie duties to help others, keep promises, repay kindness,... • No ranking among these • Highly defeasible conclusions, e.g., steal to feed kids Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

16. Obligations Obligations are • For deontological theories, those that are impermissible to omit • For teleological theories, those that most promote good • For contract-based theories, those that an agent accepts Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

17. Asimov’s Laws of Robotics 0. A robot may not injure humanity or, through inaction, allow humanity to come to harm. [Added after the following more famous laws] 1. A robot may not injure a human being, or, through inaction, allow a human being to come to harm. 2. A robot must obey the orders given it by human beings except where such orders would conflict with the First Law. 3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law. Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

18. Applying Ethics: 1 The ethical theories are theories • Of justification • Not of deliberation • An agent can decide what basic “value system” to use under any approach Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

19. Applying Ethics: 2 • The deontological theories (“right”) • Are narrower • Ignore practical consideration • But are only meant as incomplete constraints (out of all the right actions, the agent can choose any) • The teleological theories (“good”) • Are broader • Include practical considerations • But leave fewer options for the agent, who must always choose the best available alternative Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

20. Applying Ethics: 3 • The ethical approaches • Are single-agent in orientation • Implicitly encode other agents • An explicitly multiagent ethics would be an interesting topic for study Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

21. An Agent Should Act • Benevolently • Seeking the welfare of others • Rationally, i.e., maximizing utility • Consistent with its model of itself (including its desires and capabilities) • Predictably • Consistent with its model of other agents’ beliefs about itself Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

22. Benevolence: “A Mattress in the Road” Who will stop to pick it up? Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

23. Information System Example:A Collective Store • Benevolent agents might contribute information they have retrieved, filtered, and refined to a collective store • Utilitarian variant: Access to the collective store is predicated on contributions to it Collective Store World Wide Web... Query Agents Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

24. Challenges and Recommendations Respect autonomy and heterogeneity • Design rules for various levels: ontologies, transactions, protocols, organizations, … • Security and trust: difficult given openness • Scalability • Quality of service understood to include user needs and application specifics • User-centered requirements analysis and design to capture key functionality in a manner that works in multiple settings Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

25. To Probe Further • IEEE Internet Computing, http://computer.org/internet • DAI-List-Request@engr.sc.edu • (International Joint Conference and Journal) Autonomous Agents and Multiagent Systems • Journal of Web Semantics • Intl. Semantic Web Conference (ISWC) • Intl. Conf. on Service-Oriented Computing (ICSOC) • Intl. Conf. on Web Services (ICWS) • World-Wide Web Conference Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns

26. Chapter 25 Summary • SOC improves our effectiveness in building large-scale systems in open environments • Because of openness, SOC systems rely upon trust among components and in dealing with people • SOC technologies support aspects of trust: progressing, but not solved • Concepts from human ethics can inspire abstractions for designing SOC systems • Ethics can help make SOC systems manageable and responsive to human needs Service-Oriented Computing: Semantics, Processes, Agents - Munindar Singh and Michael Huhns