RaD-AI 2023
Program
The workshop took place on May 30, 2023, as part of AAMAS 2023
Times are shown in GMT
8:20 Opening remarks
8:30 Global objectives (discussing "Adversarial")
Quantitative Planning with Action Deception in Concurrent Stochastic Games (AAMAS accepted paper)
Chongyang Shi, Shuo Han, and Jie Fu
Should my agent lie for me? A study on humans’ attitudes towards deceptive AI (AAMAS accepted paper)
Stefan Sarkadi, Peidong Mei, and Edmond Awad
Certifiably Robust Policy Learning against Adversarial Multi-Agent Communication (ICLR accepted paper) Link
Yanchao Sun, Ruijie Zheng, Parisa Hassanzadeh, Yongyuan Liang, Soheil Feizi, Sumitra Ganesh and Furong Huang
9:00 Liz Sonenberg: Mind the gap: Towards Rational Imperfection via `Imperfect’ Rationality (Invited talk)
I will reflect on requirements and computational mechanisms for designing agents with Theory of Mind capabilities that are to be involved in human-agent interactions.
10:15 Coffee Break
10:45 Local objectives (discussing "Alignment")
Goal Alignment: Re-analyzing Value Alignment Problems Using Human-Aware AI (AAMAS accepted paper) Link
Malek Mechergui and Sarath Sreedharan
Intelligent Disobedience: A Novel Approach for Preventing Human Induced Interaction Failures in Robot Teleoperation (HRI accepted paper)
Kavyaa Somasundaram, Andrey Kiselev, and Amy Loutfi
Stubborn: An Environment for Evaluating Stubbornness between Agents with Aligned Incentives Link
Ram Rachum, Yonatan Nakar and Reuth Mirsky
11:45 Plan recognition (discussing "Attention of agents")
Attention! A Dynamic Epistemic Logic Model for Inattentive Agents (AAMAS accepted paper) Link
Gaia Belardinelli and Thomas Bolander
Exploring the Cost of Interruptions in Human-Robot Teaming Link
Swathi Mannem, William Macke, Peter Stone, and Reuth Mirsky
12:30 Lunch Break
14:00 Joel Leibo: Conformity to Social Norms (Invited talk)
When does it make sense to conform to social norms? Why do social norms form in the first place? Why are they sometimes clearly useful, e.g. norms that encourage cooperation, and other times apparently pointless e.g. taboos prohibiting the eating of certain foods? In this talk I will describe a few recent studies using a computational model of normativity based on multi-agent reinforcement learning. This line of work answers some of these questions and sheds new light on the social meaning of rebellion and disobedience in the sense of this workshop.
15:00 Consistency check (discussing "Revision")
Agent-directed runtime norm synthesis (AAMAS accepted paper)
Andreasa Morris Martin, Marina De Vos, Julian Padget, and Oliver Ray
Towards An Ethical Rebellion System Link
Ursula Addison, Matthew Molineaux and Othalia Larue
15:45 Coffee Break
16:30 Matthias Scheutz: We don’t need no... rebel robots (Invited talk)
In this presentation, I will argue that we don't have nor want rebel robots, but instead need robots that have at least a rudimentary understanding of our normative expectations and can use them to determine whether human instructions should be rejected for normative reasons.
17:30 Mediation (discussing "Explainability")
Trusting artificial agents: communication trumps performance (AAMAS accepted paper)
Marin Le Guillou, Laurent Prévot and Bruno Berberian
Beyond Rejection Justification: the Case for Constructive Elaborations to Command Rejections by Autonomous Agents
Gordon Briggs
18:15 Wrap up session
Liz Sonenberg
Professor Liz Sonenberg is a Professor in the School of Computing and Information Systems at The University of Melbourne, Australia
Her research focus, with colleagues nationally and internationally, has been on foundations of teamwork in artificial intelligence (AI), especially mechanisms to support decision making in hybrid teams comprised of humans and software agents.
She holds the Chancellery roles of Pro Vice Chancellor Research Systems and Pro Vice Chancellor Digital & Data, and is active in teaching and research in the Melbourne School of Engineering. Previously, at the University of Melbourne she has been Head of the Department of Information Systems and Dean of the Faculty of Science.
Joel Leibo
Dr Joel Leibo is a senior staff research scientist at DeepMind. He obtained his PhD from MIT where he studied computational neuroscience and machine learning with Tomaso Poggio. Joel was one of the first researchers to join DeepMind, starting as an intern in 2010, and then joining full time after finishing his PhD in 2013.
He is interested in reverse engineering human biological and cultural evolution to inform the development of artificial intelligence that is simultaneously human-like and human-compatible. In particular, Joel believes cooperation is the quintessential human ability.
Matthias Scheutz
Professor Matthias Scheutz is the Karol Family Applied Technology Professor of cognitive and computer science, director of the Human-Robot Interaction Laboratory and director of the human-robot interaction degree programs at Tufts University.
His current research focuses on complex ethical cognitive robots with natural language interaction, problem-solving, and instruction-based learning capabilities in open worlds.
He has over 400 peer-reviewed publications in artificial intelligence, artificial life, agent-based computing, natural language understanding, cognitive modeling, robotics, human-robot interaction and foundations of cognitive science.
Organizing Committee
David Aha
Navy Center for Applied Research in AI
Naval Research Laboratory
Washington, DC; USA
Gordon Briggs
Navy Center for Applied Research in AI
Naval Research Laboratory
Washington, DC; USA
Reuth Mirsky
Dept. of Computer Science
Bar Ilan University
Israel
Ram Rachum
Dept. of Computer Science
Bar Ilan University
Israel
Kantwon Rogers
Dept. of Computer Science
Georgia Tech
USA
Peter Stone
Dept. of Computer Science
University of Texas at Austin
TX, USA
Sony AI