Philippe Collard

Artificial Life (2024) 30 (2): 171–192.

This article deals with individuals moving in procession in real and artificial societies. A procession is a minimal form of society in which individual behavior is to go in a given direction and the organization is structured by the knowledge of the one ahead. This simple form of grouping is common in the living world, and, among humans, procession is a very circumscribed social activity whose origins are certainly very remote. This type of organization falls under microsociology, where the focus is on the study of direct interactions between individuals within small groups. In this article, we focus on the particular case of pine tree processionary caterpillars (Thaumetopoea pityocampa). In the first part, we propose a formal definition of the concept of procession and compare field experiments conducted by entomologists with agent-based simulations to study real caterpillars’ processionaries as they are. In the second part, we explore the life of caterpillars as they could be. First, by extending the model beyond reality, we can explain why real processionary caterpillars behave as they do. Then we report on field experiments on the behavior of real caterpillars artificially forced to follow a circular procession; these experiments confirm that each caterpillar can either be the leader of the procession or follow the one in front of it. In the third part, by allowing variations in the speed of movement on an artificial circular procession, computational simulations allow us to observe the emergence of unexpected mobile spatial structures built from regular polygonal shapes where chaotic movements and well-ordered forms are intimately linked. This confirms once again that simple rules can have complex consequences.

Read the full article at: direct.mit.edu

Jairo F. Gudiño , Umberto Grandi and César Hidalgo

Roy Soc Phil Trans A Volume 382I ssue 2285

We explore an augmented democracy system built on off-the-shelf large language models (LLMs) fine-tuned to augment data on citizens’ preferences elicited over policies extracted from the government programmes of the two main candidates of Brazil’s 2022 presidential election. We use a train-test cross-validation set-up to estimate the accuracy with which the LLMs predict both: a subject’s individual political choices and the aggregate preferences of the full sample of participants. At the individual level, we find that LLMs predict out of sample preferences more accurately than a ‘bundle rule’, which would assume that citizens always vote for the proposals of the candidate aligned with their self-reported political orientation. At the population level, we show that a probabilistic sample augmented by an LLM provides a more accurate estimate of the aggregate preferences of a population than the non-augmented probabilistic sample alone. Together, these results indicate that policy preference data augmented using LLMs can capture nuances that transcend party lines and represents a promising avenue of research for data augmentation.

Read the full article at: royalsocietypublishing.org