New scientific discoveries powered by artificial intelligence: Award-winning teams reimagine the future of research at the AI Hackathon

For the second consecutive year, HUN-REN organized its AI Hackathon. This year, ten teams competed, once again presenting a wide range of exciting and diverse ideas. Some teams proposed solutions aimed at reforming the scientific research process itself, while others focused on using artificial intelligence to make the evaluation of academic publications more transparent.

The competition was evaluated by a distinguished professional jury chaired by Roland Jakab, CEO of HUN-REN. Jury members included András Benczúr, Head of the Artificial Intelligence Research Laboratory at HUN-REN SZTAKI; Dr Zsolt Szalay, Associate Professor and Head of the Department of Vehicle Technology at Budapest University of Technology and Economics; Dr Gábor Kertész, Vice Dean of the John von Neumann Faculty of Informatics at Óbuda University; and Levente Szabados, an artificial intelligence expert. In their evaluation, the jury assessed not only the level of innovation behind the teams’ ideas, but also their practical feasibility.

In the end, the jury selected the idea proposed by the GeNEX team as the winner. The team put forward an autonomous AI system that breaks with the conventional “step-by-step” research approach and instead seeks out new, unexpected connections between distant scientific fields. The system analyses the abstracts of scientific publications and deliberately draws on research areas that are usually considered separate, combining them in ways that are more likely to lead to breakthrough hypotheses. A learning algorithm within the system helps determine which connections are genuinely feasible, reducing random guesswork and increasing the chances of high-impact discoveries.

The second-place team, Team Nexus, developed a system that links every claim to verifiable references and builds a provenance graph to address misinformation and gaps in knowledge. The third-place team - AI Based E2E Research Lab Automation - proposed an approach that treats laboratory work as a cyclical learning process, enabling faster and more efficient research.

In addition, the jury - through a prize offered by Óbuda University - also recognized the most creative idea. The Highway2Health team envisioned a science-based, AI-supported preventive healthcare platform for older adults, aimed at enabling safe lifestyle changes, risk screening, and personalized coaching. Their goal is to create medically supervised, adaptive guidance to support a longer, healthier, and more active life.

“Once again, the participants delivered exemplary performances. A number of ideas and prototypes emerged that are well worth taking forward. We hope that several of the concepts that took shape here will soon become technologies available to the entire research network, technologies capable of enabling or radically accelerating scientific breakthroughs,” said Roland Jakab, Chair of the Jury, at the awards ceremony.

The top three teams were invited to the HUN-REN AI Symposium 2026, while members of the winning team will also have the opportunity to attend the international Helmholtz AI Conference. In addition, thanks to HUN-REN, the three winning teams will receive automatic “open-ended” access to the research network’s AI resources, while the remaining teams may apply for access through a competitive process.

Artificial intelligence is more than just a tool. According to CEO Roland Jakab, the results clearly demonstrated that through collective thinking, a clear vision, and consistent effort, the foundations for genuine scientific breakthroughs can be established. “At HUN-REN, we do not view artificial intelligence merely as a tool, but as a research infrastructure capable of accelerating and fundamentally transforming the process of scientific discovery,” he said.

Hackathon participants applied the principles of so-called Agentic Discovery, which enable the coordinated learning and tool use of autonomous systems operating under human supervision. “These principles go beyond a single competition; they reflect the mindset through which we aim to integrate artificial intelligence into HUN-REN’s everyday research practice,” the CEO added.

HUN-REN’s long-term objective is to create an AI-ready research environment across the entire network that is transparent, accountable, and genuinely supportive of research processes. To this end, the research network plans to develop practical AI services and shared platforms that researchers can use on a daily basis, enabling ideas to evolve into internationally recognized results.

A further goal is for artificial intelligence to take on the role of an active, integrated “co-researcher” in scientific work, while strengthening Hungary’s long-term innovation performance through the development of shared research capacities across the network.