2nd International Workshop on Deceptive AI @IJCAI2021 (Yellow1 in Gathertown!!)

Program

ALL SCHEDULE TIMES GIVEN IN BST (LONDON TIME)

Session 1 UK/OZ (London from 8am/Melbourne from 5pm)

Montreal times: Thursday 19th, 3am-7am

8am Panel 1:

What are the current challenges and roadblocks to effective deceptive AI?
Discussed by: Cristiano Castelfranchi, Carlo Kopp, Chiaki Sakama, Liz Sonenberg

9am Paper presentations:

Full - Learning to Deceive in Multi-Agent Hidden Role Games (Matthew Aitchison, Lyndon Benke and Penny Sweetser) - PDF

Short - Modelling Strategic Deceptive Planning in Adversarial Multi-Agent Systems (Lyndon Benke, Michael Papasimeon and Tim Miller) - PDF

Full - Deception in Epistemic Causal Logic (Chiaki Sakama) - PDF

10am Panel 2:

Is it possible to establish a best practice for the design & deployment of deceptive agents?

Discussed by: Martin Caminada, Timotheus Kampik, Sarah Keren, Marija Slavkovik

11am Invited talk: Carlo Kopp - Understanding the Science of Deception

Abstract: As the bandwidth of digital media has grown exponentially over the last two decades, deception has become a globalised pandemic problem presenting challenges in every area of social interaction. Science research exploring how deception works has lagged empirical non-scientific research on deception badly, mostly as deception is a challenging problem spanning several distinct problem areas including information theory, decision theory, game theory, and cognitive cycles. This presentation will briefly present known challenges in deception research, deception effects, and key concepts in scientific modelling of deception using information theory, decision theory, game theory, and cognitive cycles. Finally, challenges in applying the theoretical models to practical problems in AI will be discussed.

Session 2 UK/US (London from 2pm/NY from 9am)

Montreal times: Thursday 19th, 9am - 1pm

2pm Invited talk: Serena Villata - Deceptive Argumentation: identification, reasoning and ethical challenges

Abstract: Argumentation is the process by which arguments are constructed and handled. Thus argumentation means that arguments are compared, evaluated in some respect and judged in order to establish whether any of them are warranted. The field of artificial argumentation is emerging as an important aspect of Artificial Intelligence research. The reason for this is based on the recognition that if we are to develop robust intelligent systems, then it is imperative that they can handle incomplete, inconsistent and deceptive information. In this talk, I will focus on the issue of identifying deceptive argumentation in debates and I will investigate the role of machine learning and reasoning methods to tackle this issue. I will then discuss the ethical challenges underlying the automatic identification and the potential generation of deceptive argumentation.

3pm Panel 3:

What deception-related issues arise when AI systems interact with humans?

Discussed by: Micah Clark, Nishanth Sastry, Alan Wagner, Ben Wright

4pm Paper Presentation,,

Short - Impostor GAN: Toward Modeling Social Media User Impersonation with Generative Adversarial Networks (Masnoon Nafees [deceased], Shimei Pan, Zhiyuan Chen and James Foulds) - PDF

Short - Deceptive Dreams: Nudging for Better Sleep (Hung-Chiao Chen) - PDF

Full - Influencing Choices by Changing Beliefs: A Logical Theory of Influence, Persuasion, and Deception (Grégory Bonnet, Christopher Leturc, Emiliano Lorini and Giovanni Sartor) - PDF

5pm Panel 4:

What is the purpose of models of deception? To represent? To predict (e.g., profiling liars)? To detect deception? …

Discussed by: Ronald Arkin, Emiliano Lorini, Stefan Sarkadi, Sara Uckelman