The innovation potential of artificial intelligence is vast – but so are the challenges. We spoke to the University of Luxembourg Rector about AI adoption.
Lena Mårtensson
As artificial intelligence (AI) evolves at lightning speed, it prompts both excitement and concern. From medicine to democracy, its impact is profound. Jens Kreisel, Rector of the University of Luxembourg, offers a nuanced perspective on how AI is shaping – and challenging – our world.
There is no doubt that AI is the buzzword of today, and many are trying to ride the wave. As an example of how overuse has diluted the term, Professor Kreisel shows an advertisement for a toothbrush which claims to be AI-driven. “To me, this is the very definition of a hype,” he remarks. This example underscores the widespread adoption – and occasional trivialisation – of the term AI, making it harder to discern genuine advancements from marketing gimmicks.
Today, we observe a strong democratisation of AI, which empowers individuals and organisations.
Jens Kreisel, University of Luxembourg
But that is not the end of the story. This year’s Nobel Prizes in chemistry and physics were awarded to work related to AI, testifying to the huge impact of this technology on scientific discovery breakthroughs. On top of this, AI is no longer reserved for a small elite. “Today, we observe a strong democratisation of AI, which empowers individuals and organisations. Every time you use ChatGPT or similar tools, you are using a powerful supercomputer, and you are taking advantage of big data,” says Professor Kreisel.
AI is nothing new. “It’s been around since the 1950s, but for a long time it sparked interest mainly among academics. Despite significant progress, AI failed to deliver on its promises. The so-called AI winter of the 1980s refers even to a period of stagnation in AI research and development with significant cuts in funding,” Professor Kreisel points out.
The leap forward came when it became interlinked with two other technologies: high-performance computing (HPC), based on very powerful machines able to solve advanced computation problems, and the rise of big data science, aimed at extracting meaningful insights from data through processing and analytics. Their convergence allowed machine learning algorithms to be trained on much larger datasets, which in turn enabled them to learn more complex patterns and make more accurate predictions. This led to the development of new machine learning algorithms, such as deep learning, which are capable of learning from massive amounts of data and making highly accurate predictions.
“Twenty years ago, people in these fields would be in separate communities and have at best few collaborations. Today, they are coming together. You can’t do generative AI and deep learning without big data or computing power. AI, data science and HPC form together a digital continuum, which has become an essential tool for society and industry.”
For long time, this development was mostly financed by public funds and developed by researchers in public institutions. The situation is now being reversed with large language models that underpin powerful AI applications such as ChatGPT being developed by industry.
“Very few academics saw ChatGPT coming,” Professor Kreisel underlines, pointing out the dominance of the private sector in the latest Time Magazine listing of the 100 most influential people in AI. “20 years ago, almost all of them would have been university researchers. Today, 98 of the people listed work in industry.” This shift in leadership from public to private sectors underscores a fundamental transformation in AI innovation, raising important questions about the accessibility and governance of this powerful technology.
On the flip side, however, is the enormous potential for collaborative innovation. “Geoffrey Hinton, one of the Physics Nobel Prize laureates, has been working part-time at the University of Toronto and part-time at Google Brain. The Chemistry Nobel Prize laureate Demis Hassabis is the CEO of Google DeepMind. Much of AI is about public-private partnerships.”
According to Professor Kreisel, the real added value of AI is its character as a key enabling technology which empowers interdisciplinary research and inter-sector innovation. “Most societal problems – the impact of the environment on our health, for instance – cannot be solved by single disciplines but require interdisciplinary approaches. At the University of Luxembourg, we have found that 80% of our interdisciplinary research projects have data, AI or high-performance computing at their heart. AI is a driver of interdisciplinarity!”
Combining AI and life sciences will be incredibly powerful.
Jens Kreisel, University of Luxembourg
This enabling technology can be used in virtually any area: finance, law, manufacturing, cybersecurity and many others. But the biggest wave to come is in life sciences and biotechnology. “Combining AI and life sciences will be incredibly powerful, from scientific discovery in medicine to innovations in healthcare.”
While the opportunities for AI-enabled innovation appear unlimited, there are also negative consequences that must be taken into consideration. “The environmental costs are significant,” Professor Kreisel points out. “The data computing processes of our digital world are very energy-consuming and are estimated to represent 2.5% of global CO2 emissions. This is expected to significantly increase in the next 10 years.” The energy demands of AI call for urgent innovations in efficiency, such as greener data centres, optimised computer chips or more efficient algorithm design.
A major risk is also AI-powered cyberattacks. For Professor Kreisel, however, the biggest danger is related to democracy. “The proliferation of fake news such as deep fakes, that are very difficult to identify, is a real issue. Fake news has always existed, but the disruptive factor today is their rapid spreading, their omnipresence in social media, and the commercial driving force behind them. We need to take this threat to our democracy seriously.”
I am not afraid that AI will take over the world – at least not yet. But transformative technologies such as AI require exceptional care and foresight.
Jens Kreisel, University of Luxembourg
Whether we are already seeing human-like intelligence is disputed. “Neuroscientists believe that the connection of our brain to a body provides a clue to understand consciousness, which is one the specificities of us humans compared to a computer. We can feel things like love, hunger, jealousy or pain. We are also experts at contextualising information and putting it in a technological, social, cultural or moral context to assess potential consequences in the short- or long-term perspective. I am not afraid that AI will take over the world – at least not yet. But as the Nobel laureate Demis Hassabis said, transformative technologies such as AI require exceptional care and foresight.”
By balancing innovation with caution, all sectors, industries and organisations can benefit from AI. Professor Kreisel still advises organisations not to expect the implementation of artificial intelligence to be a quick fix or a magic solution. Sharing some lessons learnt from the University of Luxembourg, he points out the following:
AI clearly holds immense promise and will enable innovation on an unprecedented scale. According to Professor Kreisel, the challenge lies in navigating both its opportunities and risks responsibly.
Photo credits: University of Luxembourg/Sophie Margue
