Information for
Quick links
Faculties
Large language models (LLMs) such as ChatGPT have become an integral part of everyday life for many people and businesses and are increasingly being used for more complex tasks, such as creating work schedules or production workflows. But in practice, it’s become clear that AI reaches its limits with such tasks – especially when information is highly similar, nested, or needs to be gathered from multiple sources. It begins to hallucinate, makes mistakes, and ultimately draws the wrong conclusions. This is precisely where the research of computational linguist Michael Hahn comes in. Together with his team in the Department of Language Science and Technology, he aims to use large datasets to investigate which relationships AI systems can infer successfully from the data, where it still falls short, and how they learn exactly – with the goal of better predicting and overall improving its reasoning.
For his research project titled REALM: Foundations for Reliable Language Model Reasoning via Inductive Biases, Michael Hahn has now been awarded an ERC Starting Grant worth €1.5 million over five years by the European Union. It is one of only five ERC Starting Grants in the field of computer science awarded to Germany in the current funding round. Following the approval of an Emmy Noether Research Group and the receipt of the Heinz Maier-Leibnitz Prize, this grant marks Michael Hahn’s third prestigious funding award or honor within a single year.
The Department of Language Science and Technology extends its warmest congratulations on this success!
The official press release from Saarland University can be read here; the Saarland Informatics Campus also reported on the news.