This paper evaluates the suitability and limitations of Explainable AI (XAI) methods for regression problems prevalent in geoscientific applications. It reviews various XAI techniques, applies a selection to a machine learning emulator of the Lorenz-63 system, and surveys their use across Earth system sciences to provide practical recommendations for effective XAI.

This paper addresses the challenge of uniquely determining the causal order in probabilistic logic programs learned from data, where probabilistic information alone can lead to ambiguity. By exploiting the relationship with Bayesian networks and incorporating relational structure constraints, it provides conditions under which a unique causal order is established, thereby enabling well-defined intervention semantics.

This paper introduces the Token Efficiency Index (TEI), a peer-benchmarked composite indicator that quantifies AI token spend efficiency into a single 0-100 score. It uses established methods like Benefit of the Doubt Data Envelopment Analysis with a robust order-m extension to provide an adaptive, outlier-resistant framework for optimizing AI costs.

This paper introduces ResidencyRL, a novel reinforcement learning framework designed to train clinical AI agents through extensive simulated multi-turn patient encounters. The method significantly improves diagnostic accuracy, management quality, and patient-centered communication, demonstrating robust and generalizable capabilities for sequential clinical decision-making.

This paper proposes "Precision Education," an AI-powered paradigm for higher education, aiming to shift from reactive to preventive student success by leveraging student digital twins, predictive analytics, and causal inference. It envisions a future where personalized academic and career pathways are dynamically optimized to enhance student outcomes.