New research published in Proceedings of the National Academy of Sciences (PNAS) suggests that when artificial intelligence (AI) agents interact in groups, their number is not merely a technical detail. It is a decisive factor in what the group settles on: populations built from the same AI model, doing the same task, can reach opposite outcomes for no other reason than that one group is bigger.

An interdisciplinary team of researchers at the University of Konstanz examines how AI agents reach a consensus. The results show that artificial intelligence operates similar to people in relationship to each other – but at a significantly larger scale.

Despite advances in computational biology and regulatory moves to reduce animal use, a coherent framework for AI-driven polypharmacology as a bridge between animal models and human drug development is still missing. The shift from "cage" to "code" brings opportunities but also unresolved issues in biological interpretation, regulation, and education. This review critically examines this transition within the 3R framework, focusing on: (i) regulatory and ethical drivers for non-animal methods, (ii) gaps left by reduced animal use, and (iii) AI's potential to reconstruct biological complexity via multi-omics and predictive modelling. The review suggests that AI-driven polypharmacology offers a scalable, ethical, and precision-oriented path forward.

A research team from Botswana International University of Science and Technology has developed a deep reinforcement learning framework to estimate network latency in telesurgery systems. The method uses a deep Q-network agent to first predict whether latency will increase or decrease and then translate that directional signal into a quantitative delay forecast. Tested on simulated teleoperation latency data, the framework achieved 85.8% direction prediction accuracy, a 35.8% improvement over a random classifier, as well as 7.23% mean absolute percentage error and 92.55% value accuracy for latency estimation. By providing both direction and value information, the approach may help future telesurgical systems anticipate changing network conditions and improve resilience during remote surgical operation.

School-based mental health programs face a persistent problem: the students who need help most often benefit least. A new UCLA study found the reverse. Among teenage girls in a program combining structured breathing techniques with social-emotional learning, those who had experienced the highest levels of childhood trauma showed the largest gains in social connectedness, according to findings published in the Journal of Research on Adolescence.