Disrupt Science Hackathon
Push the boundaries of AI-first scientific discovery
Disrupt Science Hackathon 2025 – Event Summary
The first-ever Disrupt Science Hackathon, organized by HUN-REN AI1Science initiative, brought together nearly 50 researchers and AI experts from across disciplines and countries to explore how artificial intelligence can radically reshape the way science is done.
Over the course of two intense days, 8 interdisciplinary teams worked side by side to develop prototype solutions for scientific challenges — ranging from social science automation to AI-assisted biodiversity tracking. Each project aimed to rethink the scientific method itself through an AI-first lens.
The event was supported by a team of international mentors and supported by OpenAI, whose solutions architects provided mentorship and provided tools during the Hackathon.
At the closing pitch session, the teams presented their ideas in 8-minute pitches, followed by Q&A with a distinguished international jury. The top three teams were invited to showcase their work at the AI Symposium 2025, taking place May 8–10 at the Várkert Bazár in Budapest.
Thank you to all participants, mentors, jury members, and partners who made this event such a success. The future of science is collaborative, AI-powered — and just getting started.
Top 3 Projects
AutoDiscover - 1st Place
AutoDiscover
A research-support agent system that uses analogical reasoning to suggest solutions to scientific problems by drawing on methods from distant, unrelated fields. The system promotes creativity and cross-domain discovery, helping researchers think beyond their usual boundaries.
Team Members:
Diána Balázsfi
Institute of Experimental Medicine (HUN-REN KOKI)
Péter Csáki-Szőke
Semmelweis University
Dániel Horváth
Institute for Computer Science and Control (HUN-REN SZTAKI)
Gábor Marton
Research Centre for Astronomy and Earth Sciences (HUN-REN CSFK)
Balázs Meszéna
Wigner Research Centre for Physics (HUN-REN Wigner)
Team AutoDiscover - Dániel Horváth, Diána Balázsfi, Balázs Meszéna, Péter Csáki-Szőke
Team AutoDiscover - Péter Csáki-Szőke, Diána Balázsfi, Dr. Richard Vécsey (mentor), Dániel Horváth, Balázs Meszéna
HippocampNet - 2nd Place
AI Children Simulation
This bold concept proposed building a simulated population of 'AI children' using language models to support behavioral and cognitive science. The goal was to create a safe and controllable environment for studying child memory using generative AI.
Team Members:
Kristóf Csorba
Budapest University of Technology and Economics
Attila Keresztes
Research Centre for Natural Sciences (HUN-REN TTK)
Hunor Kis
Research Centre for Natural Sciences (HUN-REN TTK)
Luca Szegletes
Budapest University of Technology and Economics
Béla Weiss
Research Centre for Natural Sciences (HUN-REN TTK)
Team HippocampNet - Kristóf Csorba, Hunor Kis, Luca Szegletes
Team HippocampNet - Luca Szegletes, Hunor Kis, Attila Keresztes
The Zeus Project - 3rd Place
AI Research Assistant for the Social Sciences
A modular system that helps automate key stages of research in the social sciences — from idea generation and narrative writing to experimental design and result analysis — using large language models and structured prompts.
Team Members:
Levente Littvay
Centre for Social Sciences (HUN-REN TK)
Klára Szalay
Parliamentary Research Service of the National Assembly (Országgyűlés)
Tamaki Eduardo Ryo
German Institute for Global and Area Studies
Team Zeus Project - Levente Littvay, Tamaki Eduardo Ryo
Team Zeus Project - Tamaki Eduardo Ryo, Klára Szalay
Meet Our Mentors
Industry leaders and academic experts will guide teams throughout the hackathon
Dr. Ali Al Shakarchi
Professor in Cybersecurity
Nottingham Trent University
Konrád Kiss
AI Strategist
AI Origo
Tom Pakeman
Solutions Architects
OpenAI
Gergő Szabó
AI4Science tech lead
HUN-REN
Dr. Richárd Vécsey
AI developer and advisor
Independent
Mitch Welzen
Solutions Architects
OpenAI
Meet Our Jury
Distinguished experts who will evaluate the hackathon projects
Dr. Vanessa Camilleri
Senior Lecturer, Artificial Intelligence
University of Malta
Dr. Gábor Kertész
Vice Dean
Óbuda University
Seyedali Mirjalili
Professor of Artificial Intelligence
Torrens University Australia
Gergely Szertics
Head of AI4Impact
HUN-REN
The Disrupt Science Hackathon is designed to push the boundaries of AI-first scientific discovery. Our goal is to bring together researchers, AI experts, and innovators to explore how AI can redefine scientific methodologies. This event fosters collaboration, innovation, and rapid experimentation in an AI-driven research environment.
Purpose of the AI1S hackathon
The Disrupt Science Hackathon aims to create an ecosystem where cutting-edge AI technologies are leveraged to tackle some of the most pressing scientific challenges. By bringing together interdisciplinary teams, we seek to accelerate AI-first research methodologies and demonstrate their transformative potential across various scientific domains.
This hackathon is structured to facilitate intensive collaboration and hands-on AI exploration. Participants will engage in a four-week preparatory phase, followed by a two-day in-person event in Budapest, culminating in a showcase at the AI1Science Symposium for the most promising projects.
Who Can Join
No AI Experience Needed
Domain scientists do not need AI expertise to participate! Our hackathon is designed to foster collaboration between domain scientists and AI specialists.
If you're a researcher in any scientific field with interesting problems and datasets, we welcome your participation. You'll be matched with AI experts who can help translate your domain knowledge into AI solutions. This is a unique opportunity to learn how AI can accelerate your research without requiring prior AI experience.
Of course, individuals who understand both domains (science and AI) are more than welcome to join and will be valuable assets to their teams.
Individual & Team Registration
Both individuals and teams can register! Whether you're coming alone or with colleagues, you'll have the opportunity to collaborate and form effective teams during the event.
Individual participants will be matched with complementary skills, while existing teams can enhance their capabilities with new members.
Hackathon Format
Three-Week Inspirational Webinar Series
Leading up to the hackathon, we will host a series of weekly inspirational talks. These sessions will feature top-tier AI researchers and scientists discussing groundbreaking applications of AI in science, preparing participants for the hands-on hackathon experience.
Two-Day Hackathon in Budapest
At the start of the hackathon, participants will have one minute to pitch their ideas, after which teams will be formed. Up to 20 teams will develop a minimum viable product (MVP) showcasing how AI can revolutionize a specific domain or research methodology and highlight its potential impact.
Teams will have access to top-tier technology mentors for guidance and will present their final results in a five-minute pitch to a panel of esteemed judges at the end of day two.
Presentation by the Top 3 Teams at the AI1Science Symposium – May 8, 2025
Exciting Opportunity! The top three teams from the hackathon will have an exclusive and prestigious chance to present their groundbreaking research at the AI1Science Symposium on May 8, 2025.
This is your chance to showcase your innovative project to leading AI and scientific experts. The symposium provides an unparalleled platform for significant exposure, networking, and potential collaboration opportunities. If your team ranks in the top 3, you'll be center stage, sharing your vision and potential impact with the scientific community.
Research Categories
The Disrupt Science Hackathon is open to researchers from ALL scientific domains. These are just examples—if your scientific field isn't listed, you can still join us!
Earth & Environmental Sciences
Use AI to understand and protect the planet: climate modeling, sustainability, geoscience, pollution tracking, biodiversity, etc.
Life Sciences & Biomedicine
AI for biology, genetics, neuroscience, medicine, pharmacology, and healthcare innovation.
Physical Sciences
Includes physics, chemistry, astronomy, and materials science—simulate, discover, or optimize physical phenomena using AI.
Mathematics & Formal Sciences
AI in pure mathematics, logic, information theory, statistics, and computational theory. Also includes AI-augmented theorem proving.
Psychology & Cognitive Science
AI exploring perception, learning, memory, language, and decision-making—both to model the brain and to enhance human cognition.
Social Sciences
Sociology, anthropology, economics, political science—use AI to understand societies, model systems, detect trends, or simulate change.
Humanities & Philosophy of Science
Apply AI to analyze texts, languages, history, ethics, or examine the role of AI itself in the evolution of science and knowledge.
Scientific Discovery & Methodology
Tools that reinvent how science is done—AI for hypothesis generation, literature synthesis, experiment design, lab automation, or open science.
Engineering & Applied Sciences
Bridge theory to practice—AI for robotics, energy systems, agriculture, civil/mechanical engineering, or any real-world application domain.
Transdisciplinary & Complex Systems
Tackle big-picture, integrative challenges—AI in systems thinking, network science, cross-domain modeling, or redefining the scientific method itself.