Marta Sabou 

Date/Time: 11.03.2026, 12:00 

Location: D2.2.094 

Abstract 

Neurosymbolic AI is a major trend in current AI research aiming to bridge two major AI techniques: sub-symbolic AI methods such as machine learning, deep learning or generative AI benefit from formally represented knowledge in symbolic AI artifacts such as ontologies, taxonomies and knowledge graphs. While these knowledge artifacts play a key role in the DNA of modern neurosymbolic AI systems, the way of creating them through the knowledge engineering process has significantly evolved with the advent of powerful generative AI techniques. In this talk, I will give examples of how GenAI techniques are used for knowledge acquisition and validation and reflect on their strengths and weaknesses. I will further present investigations of the best ways to combine human and AI components during knowledge engineering. Taking all these advances as a backdrop, I will explore what these mean for revising knowledge engineering methodologies to meet the new context of Neurosymbolic AI systems.