Rational verification: game-theoretic verification of multi-agent systems

Alessandro Abate,Lewis Hammond,Julian Gutierrez,Paul Harrenstein,Giuseppe Perelli,Muhammad Najib,Michael Wooldridge,Marta Kwiatkowska,Thomas Steeples

doi:10.1007/s10489-021-02658-y

Alessandro Abate, Lewis Hammond + Show 7 more

Open Access

https://doi.org/10.1007/s10489-021-02658-y

Copy DOI

Abstract

We provide a survey of the state of the art of rational verification: the problem of checking whether a given temporal logic formula ϕ is satisfied in some or all game-theoretic equilibria of a multi-agent system – that is, whether the system will exhibit the behavior ϕ represents under the assumption that agents within the system act rationally in pursuit of their preferences. After motivating and introducing the overall framework of rational verification, we discuss key results obtained in the past few years as well as relevant related work in logic, AI, and computer science.

Highlights

The deployment of AI technologies in a wide range of application areas over the past decade has brought the problem of verifying such systems into sharp focus
While synthesis problems have been increasingly studied within the verification community, rational verification has come to prominence only in the past few years
Rational verification is a recent approach to the automated verification of multi-agent systems

Summary

Introduction

The deployment of AI technologies in a wide range of application areas over the past decade has brought the problem of verifying such systems into sharp focus. The verification process reduces to establishing whether the specification formula is satisfied in the given Kripke. At the time of writing, software agents are ubiquitous: we have software agents in our phone (e.g., Siri), processing requests online, automatically trading in global markets, controlling complex navigation systems (e.g., those in selfdriving cars), and even carrying out tasks on our behalf at home (e.g., Alexa). These agents do not work in isolation: they may interact with humans or with other software agents. The field of multi-agent systems is concerned with understanding and engineering systems that have these characteristics

Objectives

Results

Conclusion