It concerns:
Interactive alignment between humans and artificial agents through proactivity of assistant professor Computer sciences Kim Baraka & dr. E. Senft (Idiap Research Institute).
Artificial agents - from voice assistants, conversational agents, interactive robots - are gradually entering our
daily lives. Despite increasing technical advances, most human-agent systems remain misaligned. Agents are often misaligned to users, which affects their adoption and trustworthiness, and users are often misaligned to agents’ true capabilities and states, resulting in poor usage. Going beyond the status quo on AI alignment research, this project proposes to look at human-agent misalignment as a two-sided knowledge alignment problem. Using psychological insights on proactivity in humans, the researchers will take an interdisciplinary approach centering interaction to align human-agent systems with a focus on robot assistants.
Machine Learning Accelerators for Science: Productive Spatial- Hardware Programming for the Scientific Community of computer scientist Tiziano De Matteis.
ML accelerators promise exceptional performance and energy efficiency for scientific computing. Yet, their full potential is hampered by a lack of comprehensive software ecosystems, making them hard for experts to use and inaccessible to scientists. This project tackles this gap by 1) equipping computing specialists with best practices, methodologies, benchmarks, and productive tools for easier development of ML accelerator-friendly applications; 2) empowering domain specialists with high-level programming tools and targeted training to simplify using these advanced devices. This project aims to democratize access to ML accelerators, reduce development complexity, and foster sustainable, high-performance applications across different scientific fields.
What is the effect of global warming on extreme weather? Lessons from shells and models from a past warmer period of climatologists Niels de Winter en Paolo Scussolini.
PAST studies how climate change affects the occurrence of storms in Europe, by looking at the past. Since weather records only go back ~200 years, researchers analyse ancient mollusc (clams, mussels and oysters) shells to detect past storms. They focus on the Last Interglacial period (127,000–118,000 years ago), a period with warm climate before the last ice age, to reconstruct storm patterns in western Europe. These findings are combined with high-resolution climate models to improve future storm predictions. Using information about past climate conditions, PAST helps us better understand how global warming influences extreme weather in the future.
Read more in the news item
AMASZINC – Analysing and MApping Strontium and Zinc IN Cremations to reconstruct diet, mobility, and funerary practices of ancient human and animal populations of earth scientist Lisette Kootker & dr. C. Snoeck (Vrije Universiteit Brussel).
For thousands of years, people have practiced cremation, leaving behind burned human and animal bones. But what can these remains really tell us? The AMASZINC project is revealing hidden stories by using cutting-edge laser technology to study the chemistry of burnt bones. By analysing zinc and strontium, we can trace where people once lived, what they ate, and even reconstruct the temperature of the cremation piles. This minimally destructive method will transform how cremation is studied, revealing new insights into ancient funerary traditions, diets, and mobility. AMASZINC will ultimately help better understand the lives of ancient populations.
Read about the other awards on the NWO website