Martin Hyben

Martin is an AI Specialist with expertise in natural language processing (NLP), machine learning, and cybernetics, specifically focusing on the early detection of disinformation campaigns, multilingual narrative analysis, and low-resource AI applications. His current primary research interests lie in developing robust AI models for supporting fact-checking, claim and narrative retrieval across diverse European languages, with a strong emphasis on user-centric evaluation frameworks leveraging state-of-the-art generative language models.

As a main contributor, Martin currently works on the DisTraceAI (Horizon Europe) project, which extends the work of vera.ai by introducing advanced text processing methods based on Large Language Models to detect manipulative campaigns in real-time, particularly within the Central European information space. He previously served as the Head of the ExU project (AI Models for Examining Multilingual Disinformation Narratives), where he led the development of cutting-edge tools for analyzing disinformation across 20+ languages. Additionally, he participated in the vera.ai consortium, responsible for check-worthy claims detection and narrative analysis, and is currently a main participant in the MEPSAI project, where he contributes to improvement of meta-research through modular prompting pipelines and automated data extraction using LLMs.

Beyond the NLP, Martin has extensive experience in applied AI and robotics. He participated in an effort to push the state of the art of General Artificial Intelligence, a general problem solving agent that can perceive, learn and adapt to the environment while improving its own structure to become a super-human intelligence. He also participated in development of the humanoid robot DiBo+, integrating NLP and AI modules, and contributed to spatial intelligence research at the University of Žilina, focusing on grid and position cells for pathfinding. His background also includes significant work in eHealth research regarding diabetes management via telemedicine and speech recognition/synthesis for low-resource embedding devices, where he received a Dean’s Award for developing novel audio features similar to MFCCs.

Previously serving as a WUDAP team coordinator for the HOPERO project within the Slovak AI European Digital Innovation Hub, Martin participated in supporting the digital transformation of Slovak companies through AI adoption. These efforts have strengthened his capacity to facilitate partnerships between academia and industry, driving practical AI adoption across the European innovation landscape.