Designing a research agenda for responsible AI-supported behaviour change

Pisa, June 10, 2025

Motivation

A prominent area of research within the Hybrid Intelligence domain pertains to Artificial Intelligence systems that support individuals in voluntarily adapting their behaviour. Such technological support can be relevant in different domains like health, sustainability and justice e.g., to help people adopt healthier lifestyle patterns, support people in making more sustainable choices, empower people in managing their chronic disease, or support victims of crimes in their healing process.

Key to developing effective behaviour support technologies is understanding why people do what they do (by learning about their motivations, habits, capabilities and needs), so that the offered support is timely and targeted at a pivotal mechanism. AI-related technologies can contribute to deepening this understanding, e.g., machine learning of observational data to gain insights in behavioural patterns, cognitive models to reason about cognitive aspects such as motivation and self-efficacy, conversational AI such as LLMs and chatbots to engage with people, or VR/AR approaches for training people or for providing visual insight into possible scenarios. 

Effectiveness, however, is not the only relevant consideration to take into account when designing behaviour change support systems. Key to developing responsible behaviour support technologies is identifying and implementing strategies for supporting individuals in their behaviour change trajectory in ways that align with core societal values such as liberty, autonomy, and (social) justice.

Especially in a time when there are many (commercial) efforts to use AI technology to subconsciously influence (individual and group) behaviour (e.g., consumer behaviour, voting behaviour), it is important to work on a research agenda that forms a counterbalancing narrative to this development, focusing on the design and development of AI technologies that users can trust by aligning with both public values and the users’ own values.

The aim of this interactive event is to develop a research agenda for the next 5 years for the field of AI-support behaviour change. Our aim is to do this in three steps. First, we will create an overview of the state of the art. Second, we will identify the most important challenges for the field. Finally, based on these two components, we aim to define a research agenda.

Agenda

9:00 Welcome

9:15 Paper presentations

Presentations of state-of-the-art on using AI for behaviour change, focusing on pressing societal issues.

  1. From Practice to Nudge: A Hybrid Intelligence Framework for Instructional Decision Support
    • Alex Liu, Min Sun, Shawon Sarkar, Victor Tian, Lief Esbenshade, Zachary Zhang and Kevin He
  2. Interactive Explanations to Resolve Misalignments in Behaviour Support Agents
    • Johanna Wolff, Victor de Boer, Dirk Heylen and M. Birna van Riemsdijk 
  3. Behaviour change support to reduce food waste
    • Senna Lindner, Jagoda Hanuszewicz and Charlotte Gerritsen
  4. Healed by Code: Hybrid Intelligence, Digital Grief, and the Ethics of Posthuman Bereavement Support-A Digital Humanities Study
    • Richa Mishra
  5. Reducing Carbon Intensity in Household Electricity Consumption through AI-Enhanced Decision Support
    • Michel Klein 
  6. Designing to Minimise Rather than Maximise Dependence on Behaviour Change Technologies
    • Lize Alberts

10:30 – 11:00 Coffee break

11:00 Last-minute pitches

We invite all participants to give short pitches about challenges that they see for AI supported behaviour change.

12:00 – 13:00 Brainstorm on challenges

We will make an inventory of the challenges for the research field that we foresee. The challenges will be grouped and will provide the input for our afternoon sessions.

13:00 – 14:30 Lunch

14.30 – 16.00 Reverse Brainstorming about solutions

During Reverse Brainstorming the challenges form the starting point of the discussion, but instead of trying to solve the problem, we will identify ways to worsen a problem. By reversing these ideas, the aim is to come up with solutions that we previously did not think of. 

16:00 – 16:30 Coffee break

16.30 – 17:15 World Café Discussions about research challenges

In smaller groups, we will discuss the outcome of the reverse brainstorm session. During this session, we have the assignment to detail the research agenda for the topic discussed at the table. We want to ask each table to write one paragraph as output of their discussion.

17:15 – 17:30 Plenary discussion about process towards vision paper.

Call for papers

We welcome two types of submissions. First, we invite participants to submit short papers (5 pages excluding references) formatted using the IOS formatting guidelines. Submissions will be done through EasyChair. Papers will be reviewed by the organizing committee and selected on the basis of their relevance to the workshop topics and their potential for encouraging fruitful discussion. Accepted papers will be published with CEUR-WS.

We also welcome extended abstracts (2 pages excluding references). Like the short papers, these submissions may be selected for lightning talks (see below), however they are not eligible for publication.

Topics of Interest

We encourage people to submit an abstract on:

  • AI approaches for understanding behaviour from sensor data;
  • AI-based techniques for deriving cognitions, preferences, or personal values from observational data;
  • Empirical studies on computer-supported coaching using reasoning and ML techniques;
  • Techniques or UI design strategies to support (perceived) autonomy;
  • The use of Explainable AI for behaviour change support;
  • Techniques for individual recommendations to encourage individuals to make positive and sustained changes;
  • AI techniques that aim to improved adherence to recommendations;
  • Methods for providing tailored and personalized feedback in a transparent way resulting in new insights.

 Important dates

  • Paper submission: Friday April 11, 2025, April 18, 2025 (extended)
  • Notification: Friday May 2, 2025
  • Workshop: Tuesday June 10, 2025

All dates are anywhere on earth.

Event format

During this event we offer an interactive format during which we, as a community, will set a research agenda for the coming years.

Authors from accepted papers will be asked to give a lightning talk in the morning session to determine the state-of-the-art of the research field. During the sessions in the afternoon we will focus on identifying challenges and potential solutions, leading to the establishment of a research agenda for the coming 5 years.

Programme Committee

  • Lize Alberts, Computer Science, University of Oxford
  • Tessa Beinema, Communication Science, VU Amsterdam
  • Willem-Paul Brinkman, Interactive Intelligence, TU Delft
  • Aart van Halteren, Social AI, VU Amsterdam
  • Marcos Oliveira, Computer Science, University of Exeter
  • Nimat Ullah, AI & Behaviour, Vrije Universiteit Amsterdam

Organisers

  • Charlotte Gerritsen, Associate Professor AI & Behaviour, Vrije Universiteit Amsterdam
  • Bart Kamphorst, Senior Researcher AI Ethics, Data School, Utrecht University
  • Michel Klein, Associate Professor AI & Behaviour, Vrije Universiteit Amsterdam