VR and AI in victim-offender mediation

In this interview, you’ll read how VR and AI can improve mediation between victims and offenders, preventing escalation of conflicts.
VR and AI in victim-offender mediation
In 2023, Charlotte Gerritsen (VU) and Sven Zebel (UT) received seed funding from the internal VU-UT call for their project “How avatar-based VR applications can radically improve victim-offender mediation uptake and effectiveness.” The project was carried out within the Secure Societies impact coalition. Recently, they submitted a request for a major follow-up project to NWO, with the outcome expected in early July.
“We are particularly interested in how VR (Virtual Reality) and AI (Artificial Intelligence) can be used in mediation between victims and offenders of crimes. In such cases, going to court often seems like the obvious step, but underlying conflicts and needs are usually not resolved, leaving the chance for further escalation and new conflicts. Mediation can be a good way to resolve these conflicts sustainably, and technology can play a significant role in this.”
Feeling heard
“We don’t just want to look at victims and offenders from a legal perspective but much more from a responsive approach to interacting with people. How do you ensure that victims and offenders feel more heard and that they can reach a better outcome together? The role of technology is very interesting, but it’s still rarely used in the mediation field. That’s why we started this project.”
Like-minded individuals
“We were both individually interested in this topic. Following the VU-UT seed funding call, we began looking for like-minded individuals and found each other. Our first contact was via email, asking, ‘Are you interested in this basic idea, and would you like to collaborate?’ Of course, we later met in person.”
Exploration stage
“Then our ‘exploration stage’ began. That might sound trivial, but that’s not our intention. It’s mostly a way of getting to know each other and seeing if the content is a good match. You need each other’s knowledge about the technology and the various domains. We brought together our expertise: Sven in restorative mediation, offender and victim behaviour, and conflict management, and Charlotte in using AI and VR environments to support or train people.”
Safe and realistic
“In our project, we’ve further developed a first prototype of a VR application for mediation, including writing scenarios based on real-life stories. Through a dialogue system, people navigate a VR scenario with an avatar representing the other party, following a kind of decision tree. This allows people to practice conversations in a safe and realistic way. There was immediate support from the field. People found it interesting and saw its added value.”
Two tracks
“Through an NWO application, we aim to take the next step by further developing and structurally implementing this technology. We are working with organisations such as Bureau Halt, Perspectief Herstelbemiddeling and Gevangenenzorg Nederland. We are following two tracks: Track one focuses on refining and expanding the current VR application’s scenarios. Track two aims to develop a highly advanced VR application using AI where the scenario is adaptive. With this, a victim or offender can have a highly realistic conversation with the other party, with the avatar responding not only to content but also to aspects such as speech style, intonation and non-verbal communication.”
Increasing chances
“The ultimate goal of the project is twofold. After practice, people are more motivated to reach out to the other party after a crime. We expect this to increase the likelihood of actual mediation between victims and offenders. And if this doesn’t work because the other party isn’t open, they can still experience part of the dialogue they would normally have had in real life via VR.”
Giving the conflict back
“In 40 to 60% of cases where mediation is offered to victims and offenders, no contact is made. That’s quite high, but we don’t believe in forcing it. The core idea of mediation is that it’s voluntary. That’s what makes it work well. You essentially give the conflict back to the people involved. This gives them the opportunity to respond to each other’s needs and in their own way, dealing with or resolving what happened in a way that’s good for them.”
Assisting further
“Besides all our other work, we really enjoy doing this. It energises us because you learn from each other and get a glimpse into other disciplines. But most importantly, we know very well why we’re doing this: to genuinely help people move forward. That’s highly motivating. We also notice this with the organisations we approach – they’re all enthusiastic and willing to contribute.”
Potential partner
“The seed funding is definitely a good initiative from VU Amsterdam and the UT. You can’t really do it without such support because you don’t just find each other easily, and you need funding to start and implement new initiatives. Our tip is: if you have an interesting idea, find a potential partner. Once you have found the right match, things can proceed smoothly. The work visits also help with this: traveling together, eating together. That creates a bond – a kind of school trip effect.”
Contact
Charlotte Gerritsen is an Associate Professor in Artificial Intelligence at the Computer Science Department of VU Amsterdam. She studied criminology and law, earned a PhD in AI and received a VIDI grant. Her research sits at the intersection of AI and behavioral sciences.
Sven Zebel is a social and forensic psychologist and works as an Associate Professor in the Psychology Department of Conflict, Risk and Safety at the UT. His research focuses on technology-driven conflict resolution methods and restorative interventions that help de-escalate and sustainably resolve serious conflicts and events between citizens. Zebel is also a special chair in mediation at the Department of Private Law at VU Amsterdam.
We know very well why we’re doing this: to genuinely help people move forward.