Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Ensuring trustworthy and ethical behaviour in intelligent logical agents

Stefania Costantini

TL;DR

The paper addresses ensuring trustworthy and ethical behavior in evolving, logic-based agents operating in open MAS. It introduces Agent-Oriented Interval LTL (A-ILTL) and Evolutionary Semantics to enable runtime self-checking, self-repair, and introspection within agents, using meta-rules (solve/solve_not) to enforce contextual ethics. It provides a practical instantiation in the DALI language and demonstrates that runtime verification can be computationally affordable and scalable relative to pure Prolog implementations. The approach complements traditional static verification by monitoring ongoing operation, enabling liveness and safety properties and enabling countermeasures when violations occur. This work advances practical machine ethics in MAS by furnishing a self-contained, logic-based toolkit for run-time monitoring and evolution-aware verification.

Abstract

Autonomous Intelligent Agents are employed in many applications upon which the life and welfare of living beings and vital social functions may depend. Therefore, agents should be trustworthy. A priori certification techniques (i.e., techniques applied prior to system's deployment) can be useful, but are not sufficient for agents that evolve, and thus modify their epistemic and belief state, and for open Multi-Agent Systems, where heterogeneous agents can join or leave the system at any stage of its operation. In this paper, we propose/refine/extend dynamic (runtime) logic-based self-checking techniques, devised in order to be able to ensure agents' trustworthy and ethical behaviour.

Ensuring trustworthy and ethical behaviour in intelligent logical agents

TL;DR

The paper addresses ensuring trustworthy and ethical behavior in evolving, logic-based agents operating in open MAS. It introduces Agent-Oriented Interval LTL (A-ILTL) and Evolutionary Semantics to enable runtime self-checking, self-repair, and introspection within agents, using meta-rules (solve/solve_not) to enforce contextual ethics. It provides a practical instantiation in the DALI language and demonstrates that runtime verification can be computationally affordable and scalable relative to pure Prolog implementations. The approach complements traditional static verification by monitoring ongoing operation, enabling liveness and safety properties and enabling countermeasures when violations occur. This work advances practical machine ethics in MAS by furnishing a self-contained, logic-based toolkit for run-time monitoring and evolution-aware verification.

Abstract

Autonomous Intelligent Agents are employed in many applications upon which the life and welfare of living beings and vital social functions may depend. Therefore, agents should be trustworthy. A priori certification techniques (i.e., techniques applied prior to system's deployment) can be useful, but are not sufficient for agents that evolve, and thus modify their epistemic and belief state, and for open Multi-Agent Systems, where heterogeneous agents can join or leave the system at any stage of its operation. In this paper, we propose/refine/extend dynamic (runtime) logic-based self-checking techniques, devised in order to be able to ensure agents' trustworthy and ethical behaviour.
Paper Structure (16 sections, 4 equations, 3 figures)

This paper contains 16 sections, 4 equations, 3 figures.

Table of Contents

  1. Introduction
  2. Background: Verification Methods for Agent Systems
  3. Background: Machine Ethics, Ethics in Agents and MAS
  4. Checking Agents' Reactive Behaviour
  5. A Logic for Checking Agent's Behaviour over Time
  6. A-ILTL
  7. A-ILTL and Evolutionary Semantics
  8. A-ILTL for Reflexive Self-Checking: Liveness and Safety Properties
  9. Evolutionary Expressions
  10. Experimental Evaluation
  11. DALI in a Nutshell
  12. Pure Prolog Code
  13. Proper DALI Code
  14. Experiments
  15. Complexity of Check and Discussion
  16. ...and 1 more sections

Figures (3)

  • Figure 1: x axis: instance size; y axis: execution time, blue bars pure Prolog green bars DALI
  • Figure 2: Interpolation average values, blue line pure Prolog green line DALI
  • Figure 3: x axis: instance size; y axis: gain (in percentage) when using DALI

Theorems & Definitions (6)

  • definition 1
  • definition 2
  • definition 3
  • definition 4: Evolutionary A-ILTL Expressions
  • definition 5
  • definition 6