AI should be researched, developed, deployed and used in light of human values. Who should decide what these values are? It has been argued that this decision should be a social choice problem. If we should just vote on it, should the majority decide? These are the questions explored in the talk.

Map of elections is a tool for visualizing a set of elections coming either from real-life or from synthetic distributions. It shows similarity between elections and can help in visualizing their properties (such us the existence of a Condorcet winner or a score according to a given voting rule). However, so far, all the maps of elections made concerned ordinal elections. In this talk I will present preliminary results regarding making a map of approval elections. In particular, I will describe two natural metrics for measuring distance between elections and I will discuss a number of different synthetic distributions of elections, as well as the features of the elections they provide. For example, we will be wondering what is a natural counterpart of the Mallows distribution in the approval setting. I will also show preliminary results visualizing similarities between approval-based multiwinner voting rules.

This talk is based on joint (and ongoing) work with Lukasz Janeczko, Martin Lackner, Arkadii Slinko, Stanisław Szufa, and Nimrod Talmon.

In this talk I will discuss several recent papers on the topic of participatory budgeting. Participatory budgeting is usually studied as an extension of multi-winner voting, where selecting a candidate results in paying a cost, and where a budget limit constrains the cost of the selected candidates. Although highly relevant for the technical analysis, this approach does not fully capture participatory budgeting processes as they happen in real life. This talk aims at presenting other models for participatory budgeting that more closely resemble these real-life processes. We will discuss the idea of a two-stage model where agents first propose some projects and then vote over the shortlisted projects. We will also have a look at what changes when the process is repeated over several years. Finally, we will have a look at how to include additional constraints on top of the budget limit: quotas over categories of projects, dependencies between the projects, ...

We study a mechanism design problem where a community of agents wishes to fund public projects via voluntary monetary contributions by the community members. An important application of this setting is donor coordination, which allows philanthropists to find an efficient and mutually agreeable distribution of their donations. We analyze various distribution rules (including the Nash product rule and the conditional utilitarian rule) and settle a long-standing open question of Bogomolnaia, Moulin, and Stong (2005) by showing that no strategyproof and efficient rule can guarantee that at least one approved project of each agent receives a positive amount of money. The proof reasons about 386 preference profiles and was obtained using a computer-aided method involving SAT solvers.

This is joint work with Florian Brandl, Dominik Peters, and Christian Stricker.