Societal and Environmental Risks of AI (SER-AI) Symposium

The Societal and Environmental Risks of AI (SER-AI) will take place on November 12, 2026 at Vrije Universiteit Amsterdam (The Netherlands). 

During this event, we will explore the societal and environmental risks of being (“ser” in Spanish) AI users. How is that AI usage is affecting democracies? What about its effects on human cognition? What are the environmental costs of using it?

We are looking forward to welcome you for an afternoon full of insights from VU researchers about the risks of AI use and current ongoing works, where you also will have the opportunity to do hands-on activities during the workshop session and to showcase your own research. 

Felienne Hermans:  AI is the future, but whose? 

Professor of Computer Science Education, Vrije Universiteit Amsterdam.

Description: What is AI? What is AI for, and what are its goals? Can we use AI for science? For education? The dominant discourse in 2026 makes it easy to believe in 2026 that AI is good, and will make our lives and jobs much easier.

In this talk Felienne Hermans, professor of Computer Science education will take you through an alternate history of AI, programming and its epistemological development. She will take you through over a century of alternative tech voices, and discuss what we can learn from that for the current state of programming and its future.

Bio: 

Felienne Hermans is professor of computer science education at Vrije Universiteit Amsterdam, and computer science teacher at Open Schoolgemeenschap Bijlmer.

At VU Amsterdam, Felienne teaches the courses Computer Science Education and AI in Education, and she conducts research on the accessibility of the digital world in a broad sense.

Her previous work was quite technical; in her dissertation, she developed algorithms to detect errors in Excel spreadsheets and created a programming language for children, Hedy, with which you can learn to program easily and in your own language.

Her more recent work is more philosophical in nature and questions how we actually define 'computer science' or 'programming' and how those choices influence the field and the software we create.

You can learn more about her here: https://www.felienne.nl/over-mij/

Emma Beauxis-Aussalet: From technical to societal challenges with assessing AI fairness and robustness

Assistant Professor, Faculty of Science, Artificial Intelligence, Vrije Universiteit Amsterdam.

Description: AI can have critical impacts on people and society, for instance by creating bias and discrimination. These impacts are difficult to assess and communicate to practitioners or end-users, and this gap worsens the ethical issues. This talk will discuss the means to make AI limitations transparent and understandable through the lens of a seemingly simple task: measuring AI errors. We will highlight key challenges with collecting representative test data, and assessing the test data as well as the test results. We will discuss the additional ethical implications that arise when choosing the error and fairness metrics to implement. Finally we will argue that the design of error and fairness assessment methods should also encompass the design of a human system of checks and balances.

Bio: Emma Beauxis-Aussalet is assistant professor of ethical computing at the Vrije Universiteit Amsterdam. Her research investigates the means to make AI systems that are safe, fair, explainable, and transparent -- for example by modelling their errors and bias. Before becoming a researcher, she acquired multidisciplinary expertise in the industry as a designer, R&D engineer, data specialist, and project leader. The interplay between design, research, and engineering is what fuels her research on the multi-faceted impacts of AI on society. She was named one of the 100 Brilliant Women in AI Ethics in 2021.

Pascal Koenig: Artificial Government and Changing Citizen-State Relations 

Assistant Professor, Faculty of Social Sciences and Humanities, Political Science and Public Administration, Vrije Universiteit Amsterdam.

Description: Generative AI systems are increasingly able to interact with humans in natural language and even emulate empathy and emotional expressions that characterize human conduct. As these systems become more agent-like they can also express social values more directly than previous forms of AI. AI systems can thus intervene into social relations in unprecedented ways. These systems – as artificial agents, assistants, companions – can interact with people on a large scale while subtly expressing certain social values and worldviews that are embedded in their design. However, not only do AI systems directly insert themselves into people’s relationships, mimicking a social presence, but they also change people’s relationship to the institutions that govern them. As institutions use AI systems this can bear on citizens’ perceptions of these institutions and of the people working in them. AI use in government entities alters the relationship between these entities and citizens, with consequences for citizens’ social perceptions of public employees. When citizens learn about public employees relying on AI in their work, citizens perceive these employees as less relatable – lower in warmth – and as less capable at their job. This effect seems to be especially pronounced for those public employees who are directly interacting with other people as part of their job.  

Bio: Pascal Koenig is an Assistant Professor of Public Administration. His research focuses on the intersection of Artificial Intelligence and Governance. Prior to joining VU Amsterdam, he worked as a fellow at Harvard University and as an advisor at the German Corporation for International Cooperation (GIZ), planning development cooperation projects in the area of digital transformation to promote sustainable development goals. He is the author of the book Understanding the Politics of Artificial Intelligence.  

Vincenzo Stoico: Quantifying AI Energy Usage at a Glance: A Rigorous Approach

Assistant Professor, Faculty of Science, Computer Science, Vrije Universiteit Amsterdam.

Description: The energy usage of AI is widely discussed but rarely measured rigorously, most claims rely on rough estimates rather than controlled experiments. In this talk, I'll show how we measure the energy usage of AI models in our lab, the Green Lab, using controlled, repeatable experimental methods. I'll share the factor that matters most for energy consumption during training and inference, and a concrete number from our measurements that changes how you should think about "efficient" AI. You'll leave knowing what rigorous energy measurement actually looks like, and a simple strategy to reduce AI's energy usage.

Bio: Vincenzo Stoico is an Assistant Professor in the Software and Sustainability (S2) Group at Vrije Universiteit Amsterdam (VU), specialized in controlled experiments to measure software energy efficiency and performance. He leads the Green Lab, where researchers study software quality through controlled experimentation, and co-coordinates the Software Engineering and Green IT specialization of VU's Master's in Computer Science. 

Sean-Kelly Palicki:  Measuring What We Compute: Towards (More) Efficient AI Use

Doctoral Researcher, Technische Universität München 

Description: Large Language Models (LLMs) are valuable tools for social science research, but their use implies environmental costs. Every cloud-based query relies on physical infrastructure that extracts and consumes energy, water, and minerals. As researchers seeking to use AI responsibly, understanding and reducing these impacts is becoming increasingly important. This workshop is aimed at researchers interested in Green AI and developing (more) efficient LLM workflows.

This interactive workshop will:
● Introduce pragmatic approaches to reduce, reuse, and recycle in the context of LLMs.
● Demystify energy use across the AI lifecycle and explore debates around embodied AI and data centers.
● Provide hands-on training in measuring and reducing energy use and emissions across a variety of tasks and models using CodeCarbon (Py/R), AI efficiency benchmarks, and tool-use agents

More information is available here.

To ensure that everyone can get out the most of this workshop, the number of participants will be limited to 25.  

Bio: Sean Hamilton-Palicki is a doctoral student at the Technical University of Munich, Germany. His research evaluates and develops computational approaches to social science research, with a focus on the intersection of data science and political communication research methods. He studies responsible LLM use and Green AI, and is the author of the TidyCarbon R package for measuring and reporting emissions.

Klervie Toczé and Lylian Siffre:  Let's talk about sufficiency! Shifting our mindset about what sustainable computing means 

Research Associate , Vrije Universiteit Amsterdam.

PhD Student, IMT Atlantique.

Description: In this workshop, the concept of sufficiency will first be introduced and discussed among the participants, with the aim to relate it to the participants' research and to identify potential opportunities and challenges. Afterwards, the participants will be divided into teams working together on applying different sufficiency-related tools and methods to a freely chosen use case (either self-defined or chosen among a list suggested by the workshop leaders).

More information is available here. 

To ensure that everyone can get out the most of this workshop, the number of participants will be limited to 25.  

Bio: Klervie Toczé is a post-doctoral researcher at the Vrije Universiteit Amsterdam. She currently works at the intersection of Computer Science and Development Economics,  on a dashboard tool for inclusive and sustainable urban gardens within the Feed4Food project. Prior to this, she completed her PhD about resource-aware edge computing at Linköping University (Sweden), in 2024. Her research interests are related to sustainable resource management for distributed systems. In particular, she is interested in orchestration for edge computing, sufficiency as a sustainability strategy, and defining, evaluating and visualizing sustainability metrics for human-relevant systems. 

Lylian Siffre is a PhD student in Computer Science at IMT Atlantique, France. His research quantifies the energy savings of Local-First software architectures compared to Cloud-First alternatives. He develops modeling tools to help software architects design, evaluate, and communicate Local-First solutions enabling data-driven eco-design decisions for sustainable software. 

Sofia Gil-Clavel:  Creating awareness of AI risks through serious games 

Research Associate, Communication Science, Faculty of Social Sciences and Humanities, Vrije Universiteit Amsterdam.

Description: People tend to use AI without knowing the risks that AI can bring to their life, society, and environment. It is necessary to create awareness of these risks in a way that has long lasting effects. It is here where serious games can play a role. Serious games are games designed to change the way people think or behave, and where there is an intent to measure the impact of that goal. In this workshop, participants will have a hands on experience on serious games design and testing.

More information is available here.

To ensure that everyone can get out the most of this workshop, the number of participants will be limited to 20.  

Bio: Dr. Sofia Gil-Clavel is a researcher and the Lab manager of the Societal Analytics Lab, Vrije University Amsterdam. Her research investigates social dynamics using big data and  computational methods. Her current projects focus on the responsible use of artificial intelligence and comparative analyses of climate change narratives between Western Europe and Latin America. For this, she uses serious games and content analysis, respectively.