he Synergy Between Education, Research, and Industry

AI, with ChatGPT and MidJourney in particular, has undeniably made a huge buzz this year. Are we really on the brink of an “AI revolution,” or are we simply witnessing the first steps of AI becoming accessible to the broader public?

Y.L.T.: With ChatGPT, AI has suddenly entered our daily lives. My first reaction was amazement — it truly is an incredible advancement! These conversational agents seem to pass the Turing test with flying colors, which is historic. Naturally, after the initial amazement come the questions: What risks are associated? What are the real limitations? But there is no doubt it feels revolutionary. ChatGPT has demonstrated its ability to generate or modify semantically rich content in a way comparable to humans, even more efficiently in terms of time.

C.M.: From my perspective, the major revolution brought by ChatGPT is the accessibility of AI. The real innovation from OpenAI is this incredibly user-friendly interface, available to everyone. Yet, we must remember that AI cannot question its own answers: unlike humans, it does not doubt. That’s why we cannot rely 100% on AI modules in current IT tools. Only humans can contextualize a response and reflect on their own actions. It’s essential to educate users about AI limitations and to caution that AI-generated outputs cannot always be used as-is without potential risks.

THE WIDESPREAD USE OF AI ACROSS SECTORS RAISES ETHICAL QUESTIONS: anomalies in training data, threats to sensitive personal information, and accountability for harms resulting from AI systems. How can we ensure these technologies are used responsibly and ethically?

Y.L.T.: Ultimately, what we all want is for humans to remain in control of AI. This aligns with the European Artificial Intelligence Act and related EU-level discussions. AI should not be feared but understood and mastered. Education and training are crucial: users must know what AI can do while recognizing their own responsibilities.

C.M.: Similarly, I worry about “homogenization of thought” from widespread ChatGPT use. If everyone constantly relies on the same tool, outputs may become uniform, trending toward an average, consensus-driven quality.

Y.L.T.: Indeed, those risks exist. First, there is the risk of bias propagation inherent in training data. Efforts are being made to mitigate this, but it’s extremely hard to eliminate entirely. Second, there’s the danger of humans relying too much on AI, potentially losing independent thinking. We must maintain control over AI. The temptation to stop thinking for ourselves or creating content autonomously due to AI’s speed and efficiency is real and must be managed.

C.M.: Users must understand AI’s current limitations. For example, ChatGPT currently has no internet access and no knowledge beyond 2021. Awareness of these limits is key. Proper use requires contextual understanding and clear rules, particularly in specialized applications like text manipulation, where results are impressive.

Y.L.T.: I fully agree. Industrial AI applications are highly specific and tailored to precise needs. Proper deployment requires deep understanding of the context and requirements.

THE UNIVERSITY RECENTLY LAUNCHED, IN PARTNERSHIP WITH OTHER EUROPEAN UNIVERSITIES, THE CYBERUS MASTER’S PROGRAM TO TRAIN A NEW GENERATION OF CYBERSECURITY EXPERTS. COMPANIES ARE SIMULTANEOUSLY IMPLEMENTING UPSKILLING AND RESKILLING PROGRAMS TO ADDRESS THE SKILLS SHORTAGE. HOW CAN EDUCATION, RESEARCH, AND INDUSTRY WORK TOGETHER TO FILL THIS GAP?

Y.L.T.: In cybersecurity, it’s crucial to understand the ecosystem’s needs to respond effectively. The synergy between education, research, and industry is a decisive combination for Luxembourg. Regarding education, the University of Luxembourg has collaborated with LIST for over fifteen years to offer a Master in Information Security Management. They have also launched the Erasmus Mundus Master’s in Cybersecurity, CYBERUS, with the University of Southern Brittany and the Free University of Brussels. Ideally, we should develop a full master’s program in Luxembourg, with a strong AI component, or even an integrated cybersecurity track from undergraduate to master’s level. Doctoral training is also vital; the SnT currently supervises nearly 200 PhD students, providing a talent pool that must continue to grow.

C.M.: CYBERUS is extremely interesting. CLUSIL, Luxembourg’s cybersecurity professionals association, supports it and works to raise its visibility within the cyber community and with businesses. Proximus NXT actively supports new Luxembourg legislation regarding shortage occupations, including cybersecurity.

Y.L.T.: Research partnerships are equally important. At SnT, we develop bilateral collaborations with companies, especially in cybersecurity — a national strategic priority. I’ve observed remarkable AI and cybersecurity skills growth among companies partnering with our researchers. These collaborations drive Luxembourg’s digital transformation. SnT interacts with a constellation of partners, addressing complex, relevant problems that companies cannot tackle alone.

C.M.: For innovation, use cases are essential. Companies provide these concrete cases. This is how we showcase Luxembourg’s unique skills internationally. For instance, in cybersecurity, we are exploring AI-generated network traffic that mimics real-world patterns to detect malicious behavior proactively. This is promising but requires further research and the expertise of SnT researchers.

WHAT CAN WE EXPECT FROM SNT IN TERMS OF INNOVATIVE RESEARCH PROJECTS IN 2024? AND WHAT IS PROXIMUS NXT’S POSITION REGARDING THE RESEARCH PRIORITIES OF THE UNIVERSITY CENTER?

Y.L.T.: SnT focuses on four strategic areas: FinTech, space, autonomous systems, and cybersecurity — the latter being the current top priority. Despite notable achievements and high-level publications, we aim to further develop cybersecurity in synergy with public and private actors in Luxembourg. We plan ambitious initiatives, including recruiting top professors and experts in AI, cybersecurity, and cyber defense.

C.M.: At Proximus NXT, we are active across all four SnT axes. Our goal is to provide innovative, high-quality services to companies. As a telecom and cloud operator, we manage vast networks and large volumes of data. Data is critical in cybersecurity, especially in Luxembourg’s open-data economy. Detecting malicious activity, anticipating attacks, and responding effectively requires adequate data. Ultimately, our investments aim to make Luxembourg a center of excellence in cybersecurity.