What is it about?

Our paper explores how to engineer explainable AI agents that can reason about other minds and that have the goal of deceiving others during social interactions. We show that by using a computationally implementable belief-desire-intention approach to model these agents, it is easy to explain in which circumstances deception happens or not, and it is intuitive to follow step-by-step the mental processes of deceptive agents to show why or why not they achieve or not their goals. The aim of our work is to better understand how the machines of the future might behave dishonestly if they were to follow similar cognitive processes to humans when they communicate.

Featured Image

Why is it important?

It is crucial to understand deceptive AI such that we can avoid its negative effects and reap its potential benefits. Hence, it is as important to be able to mitigate or prevent its malicious behaviour, as is to be able to design future AI agents that can use deception to smoothen social interactions.

Read the Original

This page is a summary of: Modelling deception using theory of mind in multi-agent systems, AI Communications, October 2019, IOS Press, DOI: 10.3233/aic-190615.
You can read the full text:



The following have contributed to this page