Mind the gaps

Google DeepMind has released WeatherNext 3, an advanced artificial intelligence model for global weather prediction that aims to improve forecast accuracy. Better weather forecasting capabilities could support more effective disaster preparedness and inform agricultural planning across One Health domains.

Citizen science is evolving beyond isolated research projects into a foundational infrastructure where participants themselves become integral to the scientific system, as explored at the 2026 Ars Electronica Festival's examination of how communities contribute to understanding human questions.

ChatIBD is a guideline-grounded artificial intelligence platform designed for inflammatory bowel disease professionals that uses retrieval-augmented generation to answer clinical questions with citations, and early deployment data across six months showed uptake among 913 users across 69 countries with 38% meeting active-use thresholds, whilst safeguards including medication verification and clinician review were implemented to mitigate hallucination risks.

BenchMIRT is a tool that evaluates what large language models are actually measuring when they perform on standard benchmarks, revealing whether they truly understand tasks or rely on surface-level pattern matching. This work is relevant to One Health applications of AI, such as disease diagnosis or ecological prediction, where understanding whether models possess genuine reasoning versus superficial pattern recognition is critical for clinical and environmental safety.

Google DeepMind's WeatherNext AI model assisted the US National Hurricane Center in making more accurate predictions of Hurricane Melissa's landfall in Jamaica, providing communities with extra warning time to prepare. The AI system demonstrates how machine learning can improve weather forecasting accuracy for severe weather events that pose significant public health and safety risks.

A photographer named Jos Avery admitted that his widely-shared black-and-white portrait images were generated by artificial intelligence rather than captured with the Nikon camera he had claimed to use. The revelation highlights how AI-generated imagery can gain substantial social media attention before the creator discloses its origin.