At big companies, knowledge management is a challenge. Knowledge comes from a collection of information – data that is arranged and ordered in an inconsistent manner, e.g., databases, but also reports, excel sheets, graphs, email conversations, or even pictures – in an associated context. Besides, more and more knowledge today also is implicitly captured and stored in AI and machine learning models which are trained on the data for decision making and content generation, e.g., large language and multimodal models. Knowledge comes from everywhere and capturing and storing this efficiently is already a challenge. For the creation of reports, plans or presentations you want to be able to search through and retrieve the desired knowledge that is hidden and scattered in the various sources. But how can you do this? The researchers in this lab focus on how humans and AI can collaborate together for knowledge management at large companies. They will be developing human-centered approaches that will involve humans in extracting, organizing and accessing stored knowledge. 
Challenges the researchers face are multifold: how to extract knowledge from generative AI that often hallucinates information -- the fact that such systems always produce an answer even though there is no answer to the query? How to integrate knowledge implicit in AI and those in knowledge bases into a unified schema? In all those tasks including knowledge extraction, synthesis, as well as knowledge linking and integration, how to involve humans in both effectively and efficiently, given especially the different viewpoints of human stakeholders? How to support domain experts in interacting with the knowledge through natural language interfaces? How to keep the information that the system uses as accurate and up to date as possible and finally to understand what it means to use an Ai driven knowledge management system, in specific contexts faced by companies. Although the researchers develop algorithms it’s the interaction between the data and the people using the systems that they want to understand building a better system.

With this new AI guided knowledge system developed for Kickstart and DSM a next step would be to apply or develop a similar system in a completely different context, the banking world for instance. Another question that arises with the development of such a management system is where do we want to store this data or to what extend do companies want to rely on technologies that are developed by others or do we want to develop or own?