Markus Breig, KIT

Artificial intelligence (AI) increasingly helps to make power grids more efficient and stable. Especially, in critical infrastructures such as the energy system, AI-supported predictions should be as precise and comprehensible as possible. The working group led by Benjamin Schäfer has, therefore, developed a new method and presented it in the journal Nature Communications: "SHAPformer" combines transformer models - known from modern language models - with explainable artificial intelligence methods and makes visible how individual factors, such as temperatures, holidays or wind forecasts, influence a forecast.

Pierre Picchetti, KIT

Synthetic nanomaterials have emerged as promising alternatives to natural enzymes for catalytic and therapeutic applications. Yet, their limited stability, aqueous compatibility, and catalytic scope impede broader utilization. Pierre Picchetti and his group have now developed biocompatible nanoparticles that provide a robust and sustainable platform for enzyme-like catalysis in water. Their activity can be switched on and off in the presence of chemical signals similar to how nature regulates enzymes. As the nanoparticles are free of metal and well tolerated by living cells, they are suitable for intracellular applications.

Susanne Benz, KIT

How biodiverse are different landscapes, and how quickly is this changing? Such simple questions are often difficult to answer in practice. The project Bio-O-Ton funded by the Federal Ministry of Research, Technology and Space, therefore, takes a new approach: “We combine audio recordings from citizen-science projects such as Dawn Chorus with high-resolution satellite data and analyze them using modern machine learning methods”, explains project leader Susanne Benz. “Our goal is to develop a method that makes ecological changes visible at an earlier stage, reliably captures trends, and supports authorities, researchers, and the public in protecting valuable habitats.”