The Knowledge Representation Learning (krAil) research group of the department of CLDH is headed by Achim Rettinger. It sees the holy grail of artificial intelligence in the alignment of machine-learned knowledge representations with human expectations. We assume a pluralistic perspective on knowledge representation that requires alignment on an individual level.

Researched Technologies: Generative AI, specifically Large Language Models, and symbolic representations, specifically Knowledge Graphs, for
•    Infusing knowledge into AI generated information.
•    Using AI to generate knowledge from information.

Researched subjects: Our primary information source is human communication, specifically on social media, which is cross-lingual and multi-modal. 
Current research goals:
•    Developing Generative Agents for simulating human capabilities and behavior, e.g. for researching opinion formation in social networks.
•    Developing aligned AI models without undesired biases, e.g., for reducing the risk of information manipulation in democratic processes.

The krAil research group is tightly connected to the research division Information Process Engineering at the FZI.