Skip to Content

Can AI help draft witness statements? A pioneering collaboration in the Netherlands says yes

Frederik Peters
Jan 6, 2025

With interest already being shown by several Dutch judicial organizations, an exciting new collaboration between Capgemini in the Netherlands, Rijksuniversiteit Groningen (RUG), and Scotty AI looks set to transform a critical area of criminal investigations. Capturing and using eyewitness statements is often a stumbling block in achieving a successful and timely prosecution. That’s all set to change as a Living Lab project gets underway to test and ultimately realize the new automated witness-taking solution, AIWitness.

In a career that’s embraced scaling start-ups, life as a politician, and the government technology (GovTech) ecosystem, it’s no surprise that Frederik Peters, engagement director at Capgemini, is hugely excited about an ongoing GovTech project with significant public sector ramifications. Here he tells us more about the substantial potential of AIWitness

Can you tell us briefly what AIWitness is?

AIWitness is a highly advanced GovTech solution that will revolutionize the way in which eyewitness evidence is captured and used across the judicial system. As its name suggests, the core of the solution is artificial intelligence (AI). But it’s much more than just the technology: it’s also about “how” the AI is used, with a careful balancing act of ethical, societal, legal, and scale-up considerations, alongside the already proven technology.

In practical technology terms, AI-witness is an AI-voice-to-text conversational solution. Beyond capturing witness statements in real time, it can also carry out first-line tasks, such as follow-up communication via phone, WhatsApp or email, all within a fully automated process on the Scotty AI platform. So, it can become a vital contact point between citizens and the police, offering communication in 140 languages.

What problem does AIWitness solve?

The team behind AIWitness began with a bold vision from the outset. With an interest in policing and the judiciary, we wanted to find a way in which the judicial process could be started on a Monday and finalized by the weekend, rather than weeks later. We quickly realized that the first obstacle preventing us reaching this goal was the witness taking procedure.

Witness taking is typically labor intensive and lengthy. Here in the Netherlands (and elsewhere, no doubt) police capacity shortages mean that it can be several weeks before statements are taken, creating a bottleneck in the end-to-end criminal justice process. In that time memories of the incident can change as people forget often crucial details. This has a detrimental impact on the quality of the statements used in prosecutions.

To give this more meaningful context, in 2021 in the Netherlands alone, 30,000 cases failed to lead to a criminal prosecution due to lack of evidence. In many instances, witness statements weren’t taken in time, so detailed information and evidence were lost, or statements weren’t even taken due to capacity shortages. Thousands of citizens who took the time to file a report with the police were left feeling that they couldn’t rely on the government when it came to safety and justice.

So, yes, we had a bold vision to fix this situation—and we felt we could achieve it with a solution built on generative AI (Gen AI). It would be a solution that minimized frustrating delays for the justice system and ensured more criminal cases got to court.

That’s the premise of AIWitness.

What expertise has come together to develop AIWitness?

It’s intended as an exciting collaboration between Capgemini in the Netherlands, Rijksuniversiteit Groningen (RUG), and Scotty AI. This blend of consulting, academia, and AI expertise is a great example of how a collaborative approach can accelerate innovation.

I have been involved with it since the outset. All three founders (Laura Peters at RUG, Reiner Bruns at Scotty AI, and myself at Capgemini) are driven by a passion for creating a safe and secure society. Laura’s background as an associate professor of criminal law and criminal procedure gives her invaluable insight into aspects of the judicial system. In turn, Reiner is a tech-entrepreneur, with in-depth technological knowledge about conversational (generative) AI.

The challenge of how to fix witness taking lit a spark amongst all three of us. This set in motion the AIWitness initiative.

What key questions are being asked in the development of AIWitness?

While Scotty AI’s technology is our starting point—we are already 85% there with this—the witness taking-process itself needs a rethink in terms of the boundaries regarding ethical, judicial and societal nuances. After all, while something might be technologically sound, would it be suitable in a judicial setting? For example:

  • What ethical perspectives need to be taken into consideration—and how can we prevent bias in the AI system?
  • Will a judge accept an AI-derived witness statement as an accurate reflection of an incident?
  • Does an AI have the emotional intelligence needed to deal with the human sensitivities of highly-charged situations?
  • Can automated witness taking technology be trusted to safeguard the privacy of all those involved?

Questions like these make it all the more important that this is a co-created solution not only involving tech innovators, but also taking on board the more nuanced societal, cultural, and judicial aspects. That’s why we’ve now taken the next step on the AIWitness journey with the launch of a Living Lab.

What does the Living Lab aim to achieve?

This is where it gets really exciting because it’s how we will push the scope of what AIWitness can achieve. The Living Lab will explore the different aspects of creating a solution that is technically, ethically, judicially, and societally accepted. Further, we will be looking at how to ensure our solution is scalable and of such high quality that it is fully accepted in criminal court cases.

It is early days, but our Living Lab will see us engaging with public sector organizations to conduct 360-degree experiments around automated witness taking. For example, we will carry out research into what type of crimes are suitable for this solution, and what ones aren’t. And we will explore and validate the use and value of emotional recognition within legal, ethical, social and technological parameters.

How is it going so far?

We have already made exciting progress in terms of how to make automated witness taking a better user experience. For example, AIWitness can take eyewitness statements in almost any language, which reduces the need to wait for interpreters to arrive on scene or at a police station. Information can also be cross-checked in real time, something that can’t be done manually. From a citizen perspective, the solution can read through written text so that any witness who might have literacy problems can confirm whether his/her statement is being captured verbatim.

What next for AIWitness?

Interest in AIWitness is growing. Representatives of the police and the Dutch Council for the Judiciary have already indicated they would like to join us.

And while we want to release AIWitness as quickly as possible, we know we first need to build an ecosystem of organizations on which our solution will have an impact. That’s what our Living Lab will achieve—because we know automated witness taking will require new processes, so we need to understand and test what those processes might be, engaging with partners in the police and judiciary throughout.

We will also be looking at how to scale the solution drawing on Capgemini’s global reach. This is something that a start-up company is often unable to achieve on its own, despite having developed a great technological solution.

Are there implications for the public sector beyond the justice system?

Yes, of course, the adoption of artificial intelligence to bolster human resources and deliver a better citizen experience in the face of staffing shortages and budgetary constraints has ramifications for the wider public sector. For example, could a similar AI to that used in AIWitness have use cases in booking GP and hospital appointments? Certainly, any number of regular processes that involve a conversation with a person and contain repetitive tasks could benefit from this technology. In fact, the options are endless.

Clearly, all that is for the future. For the immediate term, we are excited to continue exploring and developing AIWitness in our Living Lab and to working with a growing ecosystem of partners from across the criminal justice system. At every step of the way is the aim of contributing directly to a safer and more just society, globally.

How can we find out more about AIWitness?

You can contact me for a chat (details below), or visit the AIWitness website: www.aitwitness.org

Author

Frederik Peters

Engagement director | Principal consultant | GovTech expert
“Govtech, which involves the use of innovative technologies by public organizations to address societal challenges, is crucial for modern governments to effectively manage disruptive innovations. By integrating advanced technologies responsibly, public organizations can enhance efficiency, transparency, and responsiveness. It’s essential to ensure that human needs, rights and values remain central to these developments.”