Two researchers from the L3S Research Center have been awarded ERC Proof of Concept Grants: Professor Sören Auer and Professor Marius Lindauer will receive funding from the European Research Council (ERC) to translate results from their ERC-funded research into practical applications. Each project is funded with €150,000.
AI assistant to make research faster and more precise
Professor Sören Auer is Professor of Data Science and Digital Libraries at Leibniz University Hannover (LUH) and Director of TIB – Leibniz Information Centre for Science and Technology and University Library. In his ERC Consolidator Grant project “ScienceGraph”, L3S and TIB have been jointly investigating AI-supported analysis of scientific publications.
His Proof of Concept project now aims to demonstrate how this research can be transformed into a usable AI research assistant. The assistant will generate answers based on verified research data, consistently reference underlying sources, assess studies and methods, and transparently communicate uncertainties.
The system will be tested in materials science and data science, for example to support advice on experimental protocols or to improve the reproducibility of scientific results.
“Within twelve months, the assistant is expected to reach a level of maturity that enables its use at universities, libraries, research institutions and in industrial research and development,” says Auer. “Our goal is a trustworthy AI research assistant that accelerates research without compromising scientific accuracy.”
AutoML as an interactive co-designer
Professor Marius Lindauer is Professor of Machine Learning at the Institute of Artificial Intelligence at LUH. Building on his ERC Starting Grant, he aims to further develop Automated Machine Learning (AutoML) into an interactive system in which human expertise and AutoML work closely together.
To fully exploit the potential of AI, new AI applications must be developed quickly and reliably. AutoML can make this process more efficient. However, today’s AutoML systems often function as rigid “black boxes”. As a result, developers can hardly integrate their domain knowledge, new ideas emerge more slowly, and sophisticated AI solutions for complex tasks are difficult to realise.
The Proof of Concept project “AutoML Co-Designer” is developing a first functional prototype for a new approach to AI development. The goal is to transform AutoML from a purely automated tool into an interactive system in which humans and machines jointly design AI solutions.
“If this concept is successful, it will establish the foundation for a new generation of AI development tools,” says Lindauer. “The Proof of Concept shows how ideas can be turned into working applications more quickly. It opens up new markets for specialised AI solutions that were previously too costly or complex to develop. In this way, the project lays the groundwork for a market-ready product that helps development teams gain advantages in the rapidly growing AI economy.”
With these two ERC Proof of Concept Grants, L3S once again underlines its strength in cutting-edge AI research and in transferring innovative methods into practical applications.
