arXiv subjects

Richard Servajean

Publications and source records attributed to Richard Servajean.

3 recordsLinked to original sources

arXiv2026

Measuring the metacognition of AI

A robust decision-making process must take into account uncertainty, especially when the choice involves inherent risks. Because artificial intelligence (AI) systems are increasingly integrated into decision-making workflows, managing uncertainty relies more and more on the metacognitive capabilities of these systems; i.e, their ability to assess the reliability of and regulate their own decisions. Hence, it is crucial to employ robust methods to measure the metacognitive abilities of AI. This paper is primarily a methodological contribution arguing for the adoption of the meta-d' framework as the gold standard for assessing the metacognitive sensitivity of AIs--the ability to generate confidence ratings that distinguish correct from incorrect responses. Moreover, we propose to leverage signal detection theory (SDT) to measure the ability of AIs to spontaneously regulate their decisions based on uncertainty and risk. To demonstrate the practical utility of these psychophysical frameworks, we conduct two series of experiments on three large language models (LLMs)--GPT-5, DeepSeek-V3.2-Exp, and Mistral-Medium-2508.

cs.AI↗

arXiv2025

Impact of complex spatial population structure on early and long-term adaptation in rugged fitness landscapes

We investigate the exploration of rugged fitness landscapes by spatially structured populations with demes on the nodes of a graph, connected by migrations. In the rare migration regime, we find that finite structures can adapt more efficiently than very large ones, especially in high-dimensional fitness landscapes. Furthermore, we show that, in most landscapes, migration asymmetries associated with some suppression of natural selection allow the population to reach higher fitness peaks first. In this sense, suppression of selection can make early adaptation more efficient. However, the time it takes to reach the first fitness peak is then increased. We also find that suppression of selection tends to enhance finite-size effects. We extend our study to frequent migrations, suggesting that our conclusions hold in this regime. We then investigate the impact of spatial structure with rare migrations on long-term evolution by studying the steady state of the population. For this, we define an effective population size for the steady-state distribution. We find that suppression of selection is associated to reduced steady-state effective population sizes, and reduced average steady-state fitnesses.

q-bio.PE↗

arXiv2022

Impact of population size on early adaptation in rugged fitness landscapes

Due to stochastic fluctuations arising from finite population size, known as genetic drift, the ability of a population to explore a rugged fitness landscape depends on its size. In the weak mutation regime, while the mean steady-state fitness increases with population size, we find that the height of the first fitness peak encountered when starting from a random genotype displays various behaviors versus population size, even among small and simple rugged landscapes. We show that the accessibility of the different fitness peaks is key to determining whether this height overall increases or decreases with population size. Furthermore, there is often a finite population size that maximizes the height of the first fitness peak encountered when starting from a random genotype. This holds across various classes of model rugged landscapes with sparse peaks, and in some experimental and experimentally-inspired ones. Thus, early adaptation in rugged fitness landscapes can be more efficient and predictable for relatively small population sizes than in the large-size limit.

q-bio.PE↗