Hungarian-style AI: from competence to impact, from impact to the future

I still remember the first time I stepped into a research laboratory as a young university student. The machines buzzed in the corner, green characters flickered on the monitors, and software had to be loaded manually from punch tape or floppy disks. Every model we ran and every piece of data we extracted was treated like treasure—computing time was expensive, and machine capacity was limited. A complex simulation could keep the lab’s supercomputer busy for days.

Today, an average laptop can complete the same task in just a few seconds—and AI-powered systems don’t just calculate: they understand, learn, and offer suggestions. They now work alongside us in fields such as healthcare, environmental protection, energy research, and cultural heritage. What we couldn’t even have imagined back then has now become part of our everyday lives. Now, decades later, on World AI Day, I can write about how Hungary can be not just a participant in this new scientific era, but one of its shapers.

Hungary’s bid to join the AI Factory Antenna Programme marks a historic step forward in domestic supercomputing technologies. The aim of the initiative is to enable Hungarian researchers and SMEs to begin developing AI applications even before the launch of Levente, Hungary’s own 20-petaflops HPC supercomputer, which is scheduled to go online in 2027. To this end, Hungary has entered into a strategic partnership with Forschungszentrum Jülich in Germany, home to Europe’s most powerful supercomputer, capable of 800 petaflops of processing power. This access will allow Hungarian stakeholders to begin training and optimising AI models using cutting-edge European high-performance infrastructure.

The AI Factory Antenna project, led by a consortium coordinated by HUN-REN, has a total budget of EUR 10 million, shared equally between the participating Hungarian institutions (including research institutes and universities) and the European EuroHPC Joint Undertaking. The programme not only offers capacity, but also builds competence—helping to ensure that Hungary becomes not just a user, but also a contributor to next-generation AI technologies.

Hungarians have a centuries-old tradition of commitment to innovation and scientific excellence—from the Alfréd Rényi Institute of Mathematics to the early advances in artificial intelligence achieved by SZTAKI. Our goal is not merely to consume the latest technologies, but to help drive the research and innovation revolution ourselves: by contributing results and making our own discoveries, we aim to shape the future.

It is in this spirit that the AI4Impact programme was launched, with the aim of ensuring that artificial intelligence is not just a tool, but a genuine partner in scientific research. We are guided by the conviction that breakthroughs come through education and leading by example. That is why we have built our programme around three strategic pillars:

1. Ambassador Network
   With the help of more than 25 AI Ambassadors, we are spreading the word about the latest methods and success stories throughout the country. These are the individuals who organise local workshops, professional forums and mentoring programmes within our institutions, providing direct support to researchers.
2. Central AI Support Team
   We have assembled a seven-strong team with diverse expertise, who can be involved at every stage of the research process—from defining the problem to publishing the results. The team’s goal is to provide the broadest possible support to as many researchers as possible. Its effectiveness is evident in the fact that, in the ten months since the programme began, we have delivered more than 50 successful projects. The variety of project requests we receive also highlights the importance of multidisciplinarity: the scope of opportunities is truly remarkable, ranging from the safe development of fusion energy and drought-resistant grain varieties to cancer research and the structured reading of Hungarian literature based on predefined criteria.
3. AI First Science (AI1S)
   True breakthroughs and future-proof research require more than simply integrating existing tools. Our ambition is to take the automation of research processes even further. That’s why we launched the AI-First Science initiative, which explores how and where AI could become a genuinely autonomous researcher—capable of generating hypotheses, designing experiments, and processing data automatically. While this may still belong to the future, it is crucial to adapt to the pace of progress—not to its current state.

While we continue to work in areas such as AI4Efficiency and AI4Innovation—for example, in document management, grant-writing support and market-entry analysis—the main focus of our programme remains on education, knowledge transfer and leading by example. This is how we make it clear to every researcher and institution: AI is not just a technological challenge, but a social and economic value we can create together.

International opening: successful AI Symposium in Budapest – advancing AI research in Hungary
This May, we hosted the international AI Symposium in Budapest, bringing together some of the world’s leading experts in artificial intelligence from nearly 30 countries. With close to 1,000 participants, the event attracted significant interest not only from the scientific community but also from the wider public. Prestigious international speakers, live demonstrations, and thought-provoking keynote talks made this encounter with the future of AI a truly memorable experience.

Facing the future: increasingly intelligent and autonomous AI systems
It is clear that in the coming years, we will see the emergence of artificial intelligences that not only assist us, but also carry out an ever-growing range of tasks independently. In this light, it is essential to prepare for how we will guide and regulate the operation of AI systems acting as ‘autonomous researchers’. What ethical frameworks and quality control mechanisms will allow us to give AI the space to formulate hypotheses, design experiments, and even publish results independently? The time will come when this question—still almost amusing today—becomes an increasingly real issue. Exploring the conditions under which this could become reality is undoubtedly part of our responsibility.

A three-tier approach: Augmented, Automated and Autonomous Intelligence
Within the HUN-REN AI4Impact programme, we therefore address all three levels separately:

• Augmented Intelligence – extending the capabilities of human researchers through better tools and more powerful analyses.
• Automated Intelligence – building virtual teams that can independently carry out background tasks such as data collection, pre-processing and reporting, whether in research, research support, or the societal embedding of research outcomes.
• Autonomous Intelligence – the long-term goal: systems capable of running their own research cycles and operating independently, radically expanding the potential of human researchers.

The potential of AI is truly revolutionary—but it is our responsibility to apply the technology wisely: to support researchers, strengthen the community, and prepare for the challenges ahead. This is how we can ensure that Hungary is not merely a follower, but a shaper of the global AI landscape.