Negative Economic Shocks and the Compliance to Social Norms

This paper is about why suffering a Negative Economic Shock, i.e. a large loss, may trigger a change in behavior. We conjecture that people trade off a concern for money with a conditional preference to follow social norms, and that suffering a shock makes the first motivation more salient, leading to more norm violation. We study this question experimentally: After administering losses on the earnings from a Real Effort Task, we elicit decisions in set of pro-social and anti-social settings. To derive our predictions, we elicit social norms separately from behavior. We find that a shock increases deviations from norms in antisocial settings — more subjects cheat, steal, and avoid retaliation, with changes that are economically large. This is in line with our prediction. The effect on trust and cooperation is instead more ambiguous. Finally, we conducted an additional experiment to study the difference between an intentional shock and a random shock in a trust game. We found that the two induce partially different effects and that victims of intentional losses are more sensible to the in-group belief. This may explain why part of the literature studying shocks in natural settings found an increase in pro-social behavior, contrary to our prediction.

Sequential Learning-based IaaS Composition

We propose a novel Infrastructure-as-a-Service composition framework that selects an optimal set of consumer requests according to the provider’s qualitative preferences on long-term service provisions. Decision variables are included in the temporal conditional preference networks to represent qualitative preferences for both short-term and long-term consumers. The global preference ranking of a set of requests is computed using a k -d tree indexing-based temporal similarity measure approach. We propose an extended three-dimensional Q-learning approach to maximize the global preference ranking. We design the on-policy-based sequential selection learning approach that applies the length of request to accept or reject requests in a composition. The proposed on-policy-based learning method reuses historical experiences or policies of sequential optimization using an agglomerative clustering approach. Experimental results prove the feasibility of the proposed framework.

Modeling and Reasoning with Preferences and Ethical Priorities in AI Systems

As AI systems make decisions that affect our lives, we must ensure that these systems operate according to the same constraints, guidelines, and ethical principles that a human would follow. Humans make many complex decisions that rely on their subjective preferences, but their decisions are usually also constrained by these ethical priorities. Hence it is essential to equip AI systems with the tools to evaluate whether or not preferences are compatible with these other priorities. In computer science, the Conditional Preference networks (CP-nets), which graphically represent conditional and qualitative preference relations, can be used to model, combine, and compare subjective preferences and ethical priorities. This chapter proposes that one can use CP-nets to measure the distance between an agent’s subjective preference and the ethical principles of the agent’s community in order to ensure that the decisions of an AI system are aligned with a given set of ethical priorities.

Distributive Justice in the Lab: Testing the Binding Role of Agreement

Lorenzo Sacconi and his coauthors have put forward the hypothesis that impartial agreements on distributive rules may generate a conditional preference for conformity. The observable effect of this preference would be compliance with fair distributive rules chosen behind a veil of ignorance, even in the absence of external coercion. This paper uses a Dictator Game with production and taking option to compare two ways in which the device of the veil of ignorance may be thought to generate a motivation for, and compliance with a fair distributive rule: individually-as a thought experiment that should work as a moral cue- and collectively-as an actual process of agreement among subjects. The main result is that actual agreement proves to be necessary for agents to be led towards a fair distributive principle and to generate a significant amount of compliance in absence of external authority. This conclusion vindicates the role of actual agreements in generating motivational power in correspondence with fair distributive rules.

Conditional Preference Network with Constraints and Uncertainty

In multi-attribute preference-based reasoning, the CP-net is a graphical model to represent user's conditional ceteris paribus (all else being equal) preference statements. This paper outlines three aspects of the CP-net. First, when a CP-net is involved with a set of hard constraints, solving the Constrained CP-net requires dominance testing which is a very expensive operation. We tackle this problem by extending the CP-net model such that dominance testing is not needed. Second, user's choices involve habitual behavior and genuine decision. The former is represented using preferences, while we introduce the notion of comfort to represent the latter. Then, we suggest an extension of the CP-net which can represent both preference and comfort. Third, preferences often come with noise and uncertainty. In this regard, we suggest the probabilistic extension of the Tradeoff-enhanced CP-net (TCP-net) model. The necessary semantics and usefulness of the extensions above are described. Finally, we outline some in-progress and future work.