Autonomy in the Attention Economy
Awarded NIAA grant for an interdisciplinary project to investigate misalignments between users' perceptions of their digital behaviour and the reality, and how this compares with …
•
1 min read
Read more
Bio
Lize Alberts is an Assistant Professor at the Vrije Universiteit Amsterdam whose research centres on building AI systems that understand users’ complex needs and long-term goals, and support them in ways that foster autonomy, integration, and enduring self-determination.
She works in the AI & Behaviour lab in the Department of Computer Science, and is affiliated with the Netherlands’ Hybrid Intelligence Centre. She holds a research associate position at the Unit for the Ethics of Technology at Stellenbosch University, and is associate editor of the Cambridge University Press Elements book series on AI Ethics & Society, in affiliation with the Leverhulme Centre for the Future of Intelligence.
Lize has a unique interdisciplinary background, having published and taught across human-centred AI, computational linguistics, behavioural design, philosophy, social anthropology, and cognitive science. She obtained her DPhil in Computer Science from the University of Oxford, focusing on the ethics of human-AI social interaction. Her master’s focused on supporting multi-modal grounded language learning in AI.
During her doctorate, Lize worked at Google as a student researcher on contextualising the alignment of social AI agents, laying the groundwork for the field now termed socioaffective alignment.
More information about her research can be found on her CV, Google Scholar, and below.
Education
D.Phil. Computer Science
University of Oxford
M.A. Philosophy | Cognitive & Computational Linguistics
Stellenbosch University
B.A. (Hons) Philosophy
Stellenbosch University, University of Bristol
B.A. Humanities
NWU Potchefstroom
Topics
Human-Centred AI
Behavioural Design & Psychology
Human-Agent Interaction Ethics
LLM Evaluation & Alignment
Computational Linguistics
Philosophy of Language & Cognition
Recent Projects
Behaviour Change Support with KG-Enhanced LLMs
Investigating the effect of integrating a knowledge graph (KG) as user 'mental model' into a large language model (LLM) based dialogue agent for personalised behaviour change …
•
1 min read
Read more
Benchmarking Personalised LLM Agent Alignment
Developed benchmark and pipeline to evaluate large language models' ability to attend to and appropriately handle user-specific safety-critical context in recommendations.
•
1 min read
Read more
Research
Lize’s core research focus is person-centred human-AI interaction design, originally developed in her doctoral thesis. More than minimising ethical violations or enhancing usability, this approach - inspired by person-centred care in biomedical ethics - aims to treat individuals as whole persons with complex, evolving needs and long-term goals, and asks how AI systems can be designed to genuinely support those goals while preserving and strengthening human agency.
Her current work spans three connected areas: personalising LLM-based dialogue agents for behavioural support; the design of behaviour change technologies that foster autonomous, lasting motivation rather than engagement-driven dependency; and the implications of generative AI writing tools for academic skill development. A consistent concern is how to support users agency and integration in how they spend their time and attention.
Teaching
Lize co-designed and coordinates three courses in the VU's BSc Artificial Intelligence programme.
Modelling Human Behaviour integrates perspectives from cognitive science, behavioural psychology, social network analysis, and computational modelling to give students a strong interdisciplinary foundation for understanding and modelling human behaviour. It covers the full model development pipeline - from the theoretical and ethical frameworks that underpin why and how we model behaviour, to practical implementation across multiple modelling tools and methods, including agent-based modelling in NetLogo, social network analysis in Python, and behavioural design principles for digital platforms. Students engage critically with the strengths, limitations, and risks of contemporary approaches to predicting and influencing behaviour, and consider the implications of applying these in AI and computing systems.
Research Design for AI guides students through the full research cycle of their final thesis project: formulating research questions, conducting literature reviews, selecting appropriate methodologies, handling data ethically, and communicating findings clearly and critically. The course runs alongside the BSc AI Project, supporting students with the timely and succesful completion of each thesis milestone.
Project Socially Aware Computing is a project-based course in which students conduct independent research to research, build and analyse an agent-based simulation of a social phenomenon involving interpersonal interaction.
Mission Statement
My research approach is human-centred in that it starts with identifying specific problems from users' own perspectives, before trying to find the most effective and responsible ways to address them, rather than starting with a tool and trying to find ways to utilise it. I am constantly seeking to challenge my existing views and assumptions by reading broadly and learning new techniques to find the most appropriate and ethical routes to addressing the issues I care about. My aim is to maximise the benefits of AI and digitisation while minimising unnecessary technological dependence and unintended risks. To do so, I integrate psychological theory and critical, big-picture thinking with mixed methods user studies, and theory-driven interaction design.