More Like this

1. Understanding Decision Subjects' Fairness Perceptions and Retention in Repeated Interactions with AI-Based Decision Systems

   https://doi.org/10.1145/3514094.3534201  

   Gemalmaz, Meric Altug ; Yin, Ming ( July 2022 , Proceedings of the 5th AAAI/ACM Conference on AI, Ethics, and Society (AIES))
2. How to Explain and Justify Almost Any Decision: Potential Pitfalls for Accountability in AI Decision-Making

   https://doi.org/10.1145/3593013.3593972  

   Zhou, Joyce ; Joachims, Thorsten ( June 2023 , ACM Conference on Fairness, Accountability and Transparency (FAccT))

   Discussion of the “right to an explanation” has been increasingly relevant because of its potential utility for auditing automated decision systems, as well as for making objections to such decisions. However, most existing work on explanations focuses on collaborative environments, where designers are motivated to implement good-faith explanations that reveal potential weaknesses of a decision system. This motivation may not hold in an auditing environment. Thus, we ask: how much could explanations be used maliciously to defend a decision system? In this paper, we demonstrate how a black-box explanation system developed to defend a black-box decision system could manipulate decision recipients or auditors into accepting an intentionally discriminatory decision model. In a case-by-case scenario where decision recipients are unable to share their cases and explanations, we find that most individual decision recipients could receive a verifiable justification, even if the decision system is intentionally discriminatory. In a system-wide scenario where every decision is shared, we find that while justifications frequently contradict each other, there is no intuitive threshold to determine if these contradictions are because of malicious justifications or because of simplicity requirements of these justifications conflicting with model behavior. We end with discussion of how system-wide metrics may be more useful than explanation systems for evaluating overall decision fairness, while explanations could be useful outside of fairness auditing. 

   more » « less
3. On p-Group Isomorphism: search-to-decision, counting-to-decision, and nilpotency class reductions via tensors

   https://doi.org/10.1145/3625308  

   Grochow, Joshua A. ; Qiao, Youming ( September 2023 , ACM Transactions on Computation Theory)

   In this paper we study some classical complexity-theoretic questions regarding GroupIsomorphism(GpI). We focus onp-groups (groups of prime power order) with oddp, which are believed to be a bottleneck case for GpI, and work in the model of matrix groups over finite fields. Our main results are as follows.

   •

   Although search-to-decision and counting-to-decision reductions have been known for over four decades for GraphIsomorphism(GI), they had remained open for GpI, explicitly asked by Arvind & Torán (Bull. EATCS, 2005). Extending methods from TensorIsomorphism(Grochow & Qiao, ITCS 2021), we show moderately exponential-time such reductions withinp-groups of class 2 and exponentp.

   •

   Despitethe widely held belief thatp-groups of class 2 and exponentpare the hardest cases of GpI, there was no reduction to these groups from any larger class of groups. Again using methods from Tensor Isomorphism (ibid.), we show the first such reduction, namely from isomorphism testing ofp-groups of “small” class and exponentpto those of class two and exponentp.

   For the first results, our main innovation is to develop linear-algebraic analogues of classical graph coloring gadgets, a key technique in studying the structural complexity ofGI. Unlike the graph coloring gadgets, which support restricting to various subgroups of the symmetric group, the problems we study require restricting to various subgroups of the general linear group, which entails significantly different and more complicated gadgets.

   The analysis of one of our gadgets relies on a classical result from group theory regarding random generation of classical groups (Kantor & Lubotzky, Geom. Dedicata, 1990). For the nilpotency class reduction, we combine a runtime analysis

   of the Lazard correspondence with TensorIsomorphism-completeness results (Grochow & Qiao, ibid.).

    

   more » « less
4. Introduction to special issue “Understanding Cognition and Decision Making by Children.” Studying decision-making in children: Challenges and opportunities

   https://doi.org/10.1016/j.jebo.2020.01.020  

   Brocas, Isabelle ; Carrillo, Juan D. ( November 2020 , Journal of Economic Behavior & Organization)

   null (Ed.)
5. Are you an exception to your favorite decision theory? Behavioral decision research is a linda problem!

   https://doi.org/10.1037/dec0000161  

   Regenwetter, Michel ; Robinson, Maria M. ; Wang, Cihang ( April 2022 , Decision)