The University of Amsterdam launched the Research Priority Area (RPA) Human(e) AI on May 15, in an event attended by over 150 researchers, staff members and students. The RPA is a multi-year program encompassing all UvA faculties aiming to synthesise ongoing work and stimulate new research on the societal consequences of the rapid development of artificial intelligence (AI) and automated decision-making (ADM) in a wide variety of societal areas.
The rector magnificus, prof. dr. ir. Karen Maex, opened the event, highlighting the unique position that the University of Amsterdam has in spearheading such an initiative. “I am proud to be a rector of a university embracing fundamental research in AI, its importance for biomedical science and treatments, the link with various industrial sectors – and equally important for the in-depth insights, the progress and excellence of science, in the various fields of the humanities, social and behavioural sciences, economics and law”, said the rector. “If artificial intelligence and automated decision-making are to reach their full potential and be successfully integrated into our societies, public support will be essential. Our focus should not only be on developing the technology and infrastructure, but – equally importantly – to identify and understand the wide-ranging opportunities and risks that come along with it. The Human(e) AI RPA answers this need.”
The day program was followed by University Professor Maarten de Rijke, director of ICAI, who highlighted the importance of interdisciplinary research and collaboration with the technology industry and with societal partners. The principal investigators of the new RPA, prof. Natali Helberger, prof. Julia Noordegraaf, prof. Sonja Smets and prof. Claes de Vreese then launched the new RPA. “Technology does not overcome us: it is up to us to shape it, and to define the conditions under which we want to implement it into our society.
To be able to do so we need research”, said prof. Helberger, while introducing the new RPA. And the research needed includes the excellent technology research, which the UvA has a long tradition on, as well as research “into the way in which AI is used, who uses it and for what reasons, into how humans and machines interact and with what effects for society, and on the legal conditions and ethical guidelines needed to maximize the potential and minimize the harm that AI can create”. To do this research, we need to join forces, said prof. Helberger: “We need to combine the excellent expertise we have here at the UvA in the area of tech development with our expertise in law, ethics, digital humanities, social sciences, economics, medicine, because these too are areas which our researchers belong to the thought leaders worldwide – and only if we combine our expertise we will be able to create the conditions for Humane AI, AI that contributes to the society that we want to live in”. In summary, the RPA Human(e) AI aims to connect, research, and share knowledge about the societal consequences of AI, while training a new generation of interdisciplinary researchers.
The first plenary panel of the day focused on AI & Society, which is one of the key research dimensions of the Human(e) AI RPA. The panelists -Maarten de Rijke, Bernhard Rieder, Claes de Vreese, Raquel Fernández- reflected on how the development of data-driven machine learning technologies affects decision-making processes. Of particular interest is the influence of AI on trust and participation in the political system, economy, and culture. The panelists emphasized the importance of collaborative research to explore such issues. Yet, this immediately also raised the concern that universities lack sufficient resources to facilitate the necessary research.
In the second panel, Natali Helberger, Ulle Endriss, Beate Roessler, and Iris van Domselaar discussed AI & Values. They observed that there is a strong need for more conceptual, normative research into the fundamental rights, societal values and ethics of automated decision making. What are and should these values be? And how can they guide automated decision making? Natali Helberger noted that the law will become more pervasive with the proliferation of artificial intelligence, providing the conditions for living peacefully together. The field of ethics, as Beate Roessler emphasized, is also crucial, as it is by no means arbitrary how we approach questions fairness, privacy, and freedom.
The third panel, featuring Max Welling, Sonja Smets, Han van der Maas, Julia Noordegraaf, Ivana Isgum, was concerned with AI & Science. The starting point of the panel was the observation that AI invites continuous critical assessment of the origin and composition of data, as well as how this technology processes data and shapes its output. The question is how to translate normative frameworks for evaluating bias, detecting presupposed notions of meaning in reasoning processes, and explaining the effect of epistemic assumptions. What methods can be developed to boost comprehensibility and explainability? The panelists also highlighted the need for developing the minimum requirements for transparency of systems and data, understanding the models, systems and processes, and getting insights into how the information flows among systems.
Call for Proposals
The launch event concluded with the announcement of a call for proposals by the RPA Human(e) AI open to all UvA researchers, and workshops to discuss potential projects. Researchers from a wide variety of disciplines and faculties participated in these sessions in the afternoon, already identifying common interests across disciplinary boundaries, pitching ideas and discussing future collaborations. The call for proposals is available at https://humane-ai.nl/call-for-proposals/