National Conference on "Governing AI" Program

On Friday 2 February, the Utrecht University focus areas Governing the Digital Society, Applied Data Science and Human-centered Artificial Intelligence are organizing an interdisciplinary conference focusing on the theme of “Governing AI”.

Governing AI

The conference program features renowned keynote speakers Tamar Sharon, Michael Fisher and Dipak Kotecha, as well as six contributions from Utrecht University researchers who will present their work through shorter pitches. The event aims to foster engaging and informed discussions on the pressing topics within the realm of AI governance.

09.30-10.00 Walk-in and registration

10.00-10.15 Word of welcome by José van Dijck

10.15-11.00 Keynote Dipak Kotecha - "Technology as a tool to improve clinical and patient management: Opportunities and pitfalls"

11.00-11.30 Coffee break

11.30-12.15 First round of three short presentations by Utrecht University scholars

12.10-13.30 Lunch break

13.30-14.15 Keynote Michael Fisher – Title t.b.a.

14.15-14.45 Break

14.45-15.30 Second round of three short presentations by Utrecht University scholars

15.30-16.00 Break

16.00-16.45 Keynote Tamar Sharon – "On the need to protect the autonomy and integrity of spheres: Towards a sphere-centric framework for the governance of digital society"

16.45-17.00 Closing remarks and conference wrap-up by Mehdi Dastani

17.00-18.00 Drinks

Keynotes

Prof Tamar Sharon will speak about the need to protect the autonomy and integrity of spheres in a lecture entitled Towards a sphere-centric framework for the governance of digital society. 

As we transition to digital society, the key societal spheres that make up society are being transformed. In the process, we may relinquish democratic control over spheres which increasingly rely on privately owned digital infrastructure for their proper functioning, while the values, forms of expertise and ends that these spheres have traditionally embodied and sought to realize are at risk of getting lost. Yet, dominant approaches for addressing digitalization risks – focusing on either data protection or competitive markets – do not seek to protect spheres. In this talk, Sharon proposes a novel, sphere-centric framework for the moral evaluation of digitalization. Sharon is affiliated to Radboud University and is Chair of the Department of Ethics and Political Philosophy and co-director of iHub, Radboud's interfaculty center for research on digitalisation and society.

Prof Dipak Kotecha's keynote is called Technology as a tool to improve clinical and patient management: Opportunities and pitfalls. 

The application of artificial intelligence and ‘big data’ analytics is impacting on every facet of society, however the potential value to enhancing healthcare is not yet fully realised. Novel technologies could lead to transformational changes in healthcare planning, delivery, cost-effectiveness and patient outcomes. In particular, to account for current and future disease burden and the rising tide of multimorbidity which is severely impacting on modern-day management strategies. The opportunities for these technological advances can only be leveraged for healthcare with a clear understanding of their limitations and pitfalls. This includes equitable access to data, development of standards to ensure high data quality, attention to privacy and data security, and an approach that brings in societal views and attitudes. Transparency and robust evaluation, as Kotecha will claim, will be the key determinants for the successful integration of artificial intelligence into routine clinical care. Kotecha leads global innovation teams to implement new technology and health data to benefit the care of patients, supported by public engagement. He is the Chief Investigator of clinical trials in the UK and internationally.

Short presentations

The Value of algorithm registers in the Netherlands by Esther Nieuwenhuizen. In the last two years, algorithm registers have emerged rapidly across the Netherlands. They come in various forms and sizes, differing significantly in the information they provide about algorithms. They all seem to share a common goal: to increase trust in algorithm usage by public organizations through transparency. But are algorithm registers a meaningful tool for transparency, or are they merely a box-ticking exercise for public organizations? In this presentation, I will address this question by providing insights into what these ARs are, exploring the drivers and barriers encountered by public organizations in adopting them, and delving into the implications associated with these registers. 

Nieuwenhuizen is PhD-candidate at the Utrecht University School of Governance. She mainly focuses on the influence of transparency on citizen trust in algorithm use. Her research is part of a project on responsible and trustworthy algorithmic policing in the Netherlands (ALGOPOL).

Machine learning paves the way for understanding complex biological processes by Dr Belén Rabaglino. The development and application of high-throughput technologies during the last decades have generated big data by globally measuring biological molecules such as the genome, transcriptome, and proteome. Analysis of omics data has mainly involved statistical tests and bioinformatics to comprehend complex biological events. Nowadays, several AI methods, including machine learning and deep learning, are also being used to mine these big data and have been shown as powerful tools to unravel biological mechanisms and pinpoint novel diagnostic and therapeutic procedures. These approaches have advanced human medicine, but their application in the veterinary sciences is less explored. Here, I will overview my work on using machine learning in omics data to understand pregnancy establishment in the cow, a species of major interest for food production, and the potential regulatory needs to address the use of AI in this area.

Rabaglino is  Assistant Professor of Ruminant Reproduction at the Department of Population Health Sciences, Faculty of Veterinary Medicine, Utrecht University. Her research explores and exploits the robust data derived from omics technologies to understand embryo competence and foetal development in the cow or other species.

Governing AI from within: a visionary proposal by Prof Federica Russo. In this talk, I present some ideas of how introducing virtue-based epistemology and ethics in the training of junior scientists may be another way of governing AI, but from within, and fostering ideas like common good or human flourishing (hopefully) and not just speed of computation, money etc.

Russo is full professor in philosophy and ethics of techno-science. She holds the Westerdijk Chair at the Freudenthal Institute, Utrecht University, and she is Honorary Professor at University College London. Her research concerns epistemological, methodological, and normative aspects that arise in the health and social sciences, with special attention to policy contexts and to the highly technologized character of these fields.

Towards responsible auditing: Privacy-aware fairness estimation of rules by Heysem Kaya. The protection of sensitive data becomes more vital, as data increases in value and potency. Furthermore, the pressure increases from regulators and society on model developers to make their AI models non-discriminatory. To boot, there is a need for interpretable, transparent AI models for high-stakes tasks. In general, measuring the fairness of any AI model requires the sensitive attributes of the individuals in the dataset, thus raising privacy concerns. In this talk, I will present a method that we proposed very recently, dubbed Privacy-Aware Fairness Estimation of Rules (PAFER), which can estimate a popularly used fairness measure (Statistical Parity) in a privacy-aware manner for interpretable models (Decision Trees), making use of a third-party legal entity that securely holds this sensitive data, guarantees privacy by adding noise to the sensitive data.

Kaya is assistant professor affiliated with the Social and Affective Computing research group at the Department of Information and Computing Sciences, Utrecht University. His research interests lie at the intersection of responsible AI and affective computing, aimed particularly for mental healthcare.

Big AI: Cloud infrastructure dependence and the industrialisation of artificial intelligence by Anne Helmond. This research explores how major technology conglomerates like Amazon, Microsoft, and Google play a central role in the “industrialization of AI.” This process involves the transition of AI from research and development to widespread practical use across various industries, resulting in new dependencies and associated investments. We refer to this convergence between AI and Big Tech as “Big AI.” Our study demonstrates how these tech giants have integrated AI into their cloud services, providing marketplaces for third-party developers and businesses to create industry-specific AI solutions, and analyzes its implications for Big AI’s governance.

Helmond is Associate Professor of Media, Data & Society at Utrecht University where she examines the processes of platformization, algorithmization, and datafication from an empirical and historical perspective.