Social Simulation Models at the Ethical Crossroads

Universality of opinions disappearing in sociophysical models of opinion dynamics: From initial multitude of opinions to ultimate consensus

Possibility of reaching a consensus in social systems with strong initial fragmentation is one of the most interesting issues in sociophysics. It is also intriguing what the dynamics of such processes is. To address those problems, we performed computer simulations using well-established models of social opinion formation, namely, the voter, Sznajd, and Latané models. We investigated opinion dynamics in cases where the initial number of opinions is very large, equal to the number of actors (the voter and Latané models) or when every second actor has their own opinion (Sznajd model), with some variations on the update schemes, lattice topologies, effective ranges of interaction, and information noise levels. For all considered models, the number of opinions assumed by the actors is finally almost always reduced to only one. However, while the voter and Latané models exhibit a power-law time decrease in the number of opinions, the Sznajd model follows a complex three-stage behavior. We also demonstrated the mean/median time of reaching the consensus scales with system size according to a power law for voter and Sznajd models, while for the Latané model, this increase is even faster. Our results show that in the studied models, the consensus is possible, provided that a long enough and model-dependent time to reach this state is available.

Social Depolarization and Diversity of Opinions-Unified ABM Framework

Most sociophysics opinion dynamics simulations assume that contacts between agents lead to greater similarity of opinions, and that there is a tendency for agents having similar opinions to group together. These mechanisms result, in many types of models, in significant polarization, understood as separation between groups of agents having conflicting opinions. The addition of inflexible agents (zealots) or mechanisms, which drive conflicting opinions even further apart, only exacerbates these polarizing processes. Using a universal mathematical framework, formulated in the language of utility functions, we present novel simulation results. They combine polarizing tendencies with mechanisms potentially favoring diverse, non-polarized environments. The simulations are aimed at answering the following question: How can non-polarized systems exist in stable configurations? The framework enables easy introduction, and study, of the effects of external "pro-diversity", and its contribution to the utility function. Specific examples presented in this paper include an extension of the classic square geometry Ising-like model, in which agents modify their opinions, and a dynamic scale-free network system with two different mechanisms promoting local diversity, where agents modify the structure of the connecting network while keeping their opinions stable. Despite the differences between these models, they show fundamental similarities in results in terms of the existence of low temperature, stable, locally and globally diverse states, i.e., states in which agents with differing opinions remain closely linked. While these results do not answer the socially relevant question of how to combat the growing polarization observed in many modern democratic societies, they open a path towards modeling polarization diminishing activities. These, in turn, could act as guidance for implementing actual depolarization social strategies.

Vanishing Opinions in Latané Model of Opinion Formation

In this paper, the results of computer simulations based on the Nowak-Szamrej-Latané model with multiple (from two to five) opinions available in the system are presented. We introduce the noise discrimination level (which says how small the clusters of agents could be considered negligible) as a quite useful quantity that allows qualitative characterization of the system. We show that depending on the introduced noise discrimination level, the range of actors' interactions (controlled indirectly by an exponent in the distance scaling function, the larger the exponent, the more influential the nearest neighbors are) and the information noise level (modeled as social temperature, which increases results in the increase in randomness in taking the opinion by the agents), the ultimate number of the opinions (measured as the number of clusters of actors sharing the same opinion in clusters greater than the noise discrimination level) may be smaller than the number of opinions available in the system. These are observed in small and large information noise limits but result in either unanimity, or polarization, or randomization of opinions.

Hammering with the telescope

The rapid pace in which various Artificial Intelligence and Machine Learning tools are developed, both within the research community and outside of it, often discourages the involved researchers from taking time to consider potential consequences and applications of the technical advances, especially the unintended ones. While there are notable exceptions to this "gold rush" tendency, individuals and groups providing careful analyses and recommendations for future actions, their adoption remains, at best, limited. This essay presents an analysis of the ethical (and not only) challenges connected with the applications of AI/ML methods in the socio-legal domain.