Background Pattern

KPN and JADS: Making AI work in practice takes more than technology

Posted on

For nearly a decade, KPN and JADS have worked together on questions at the intersection of AI research and practice. What began with two PhD projects has developed into a broader partnership spanning research, education and talent development. Gianluigi Bardelloni, Research Lead AI at KPN, reflects on how the collaboration has evolved, what it has delivered in practice, and why the connection between academia and industry matters.

A partnership from the very beginning

KPN has been involved with JADS almost since day one. The collaboration began in 2016, when KPN invested in two PhD projects on a topic that was still largely unexplored at the time: AI explainability.

Gianluigi recalls: “Back then, there were still many open questions. If an AI model makes a decision that affects someone,  should you need to explain why? What are the legal requirements? What are the ethical implications? We were really pioneering in this field.”

These first projects laid the foundation for a collaboration that continues to this day.

Growing from a few PhD projects into a broad partnership

Over the years, the collaboration grew well beyond those first two projects. Today, it includes:

  • PhD research
  • EngD projects
  • Master’s thesis projects
  • Research collaborations
  • Participation in the ICAI (Innovation Center for Artificial Intelligence) ecosystem via the ICAI Lab for Responsible AI

One of the early PhD candidates even joined KPN permanently. “He was doing excellent work, and we offered him a position, when he decided to quit the PhD. In that sense, the collaboration has also been a valuable talent pipeline.” Today, multiple PhD and EngD projects continue to run jointly between KPN and JADS.

Responsible AI as a strategic priority

For KPN, the strategic value of the partnership lies in connecting cutting-edge academic research with real-world business challenges. “Our goal is to apply state-of-the-art research to practical use cases. Many of our challenges are at the frontier of what is technically possible.” Responsible AI has been an important focus within this collaboration, including topics such as explainability, fairness and privacy.

As AI systems became more prominent, KPN and JADS explored how organizations can use them responsibly and transparently. “It is important that users understand why a model makes a recommendation or generates an answer. They should be able to assess whether it is a good prediction, a fair decision, or correctly generated content.”

From research to real-world deployment

The partnership has resulted in more than academic insights. Some research outcomes have found their way into analytical systems.

One example is the development of privacy-preserving technologies that help identify and protect personal data during processing. Another area of research has focused on knowledge graphs, which can improve AI-generated answers and reduce hallucinations. JADS EngD trainees explored how AI systems can retrieve the right information before generating a response.

“The challenge is not only generating an answer, but finding the correct information first. Knowledge graph technologies help AI systems identify the most relevant information and provide more reliable responses.”

These innovations are expected to improve the KPN AI-powered support tools for employees and customer service teams.

Talent as a catalyst for innovation

Students and researchers continue to play an important role in the collaboration. According to Gianluigi, they bring fresh ideas, current academic knowledge, and the freedom to explore new directions. “Students are often working with the newest technologies. We bring the use cases, the data and the business challenges. Together, that creates something valuable.”

The partnership also creates opportunities to attract and develop talent, with several students moving on to roles in industry.

What makes JADS different?

KPN works with multiple universities and research institutions, but Gianluigi highlights one characteristic that makes JADS stand out:

“The strength of JADS is its accessibility and its small scale. The lines of communication can be very short. Often, I can directly connect with professors and researchers to discuss ideas, challenges or opportunities. Continuing to strengthen these direct connections is an important opportunity for JADS as it develops further.”

It is just such close connection between academia and industry that creates an environment where ideas can move quickly from concept to application.

For KPN, the combination of academic expertise, practical relevance and direct collaboration continues to make JADS a valuable partner in exploring how AI can be applied responsibly in practice.

Group 5
Group 6
Group 7
JADS