Jacco van Ossenbruggen appointed Full Professor of Human-Centred Data Science
Effective 1 June, Jacco van Ossenbruggen has been appointed Full Professor of Human-Centred Data Science in the Computer Science Department at VU Amsterdam. Van Ossenbruggen will be focussing on data-driven research and data-analysis, in collaboration with researchers within and outside the university.
05/31/2021 | 1:24 PM
Van Ossenbruggen’s chair will focus on data-intensive applications in which people play a crucial role. For example, the data that are used might directly concern people, a human assessment might be essential in the analysis of the data, or people might need to interpret the results of the analyses in the right way.
Van Ossenbruggen: 'Our research always touches upon areas of application in which the volume of data is too great to process manually and in which the problem or the data are too complex or subjective to leave the analysis completely to computers. In order to arrive at good solutions in such cases, it is necessary to design the ‘human role’ from the very beginning.'
Reducing social inequality
One good example is the study conducted in the recently established Civic AI lab, in which VU Amsterdam is collaborating with the University of Amsterdam, the City of Amsterdam and the Ministry of the Interior to investigate the use of artificial intelligence (AI) within the government. By having a good understanding of the problem and the data from the beginning, Civic AI aims to design algorithms aimed at reducing social inequality. This is because many AI algorithms work surprisingly well in the lab for many different types of data. Van Ossenbruggen: 'For computer scientists, it can be tempting to abstract from the details in the data, because they “don’t matter anyway”. This ceases to be a good attitude, however, whenever the data concern people or whenever such algorithms lead to decisions that affect people. The enthusiasm for an algorithm that scores 96% correctly in the lab will evaporate quickly if, in practice, the 4% of errors that are found systematically lead to unfavourable outcomes, like greater social inequality in a given community.'
Algorithms that recognise prejudices
Another example is the design of AI systems in the Cultural AI lab, where VU Amsterdam is collaborating with partners including the Rijksmuseum, the National Library of the Netherlands and the Netherlands Institute for Sound and Vision. Because AI systems learn from data, they often unintentionally learn the undesirable patterns contained within the data (e.g. racist or sexist prejudices). Van Ossenbruggen: 'In the Netherlands, a large amount of heritage has been digitized by museums, libraries and archives, and these institutes and humanities scholars possess a large amount of knowledge about this heritage. These experts are well aware of the prejudices that are contained in their data. How can we use this knowledge to train algorithms to recognise these prejudices? Could AI systems learn that stories about the same objects or events can be told from within a wide range of different perspectives?'
Jacco van Ossenbruggen received his PhD in computer science from VU Amsterdam in 2001. Until recently, he headed a research group at the national research institute for mathematics and computer science (Centrum Wiskunde & Informatica/CWI) in Amsterdam. Since 2018, he has also headed the User-centric Data Science group at VU Amsterdam. He is one of the founders of the Civic AI and Cultural AIICAI labs. As a member of the management board of ODISSEI and CLARIAH, Van Ossenbruggen is working to improve the digital infrastructure for researchers in the social sciences and humanities.