AI in transition from LLMs to agents

Test copy of the proposal on semantic-mediawiki.org, used to test Template:AgendaDay26. Time slot and status are test values from the draft program.

AI in transition from LLMs to agents (Keynote, 45 min, In-person, Accepted)

Speakers: Marko Robnik-Šikonja · Start: 2026/11/11 11:20:00 · Conference: MediaWiki Users and Developers Conference Fall 2026

This talk explains the shift from LLMs to agentic AI, how the two differ, and why agents often use wikis for memory and coordination. It presents a system answering grammar questions from Universal Dependencies data for 180+ languages. Currently, we are observing a transition in major artificial intelligence (AI) developments from large language models (LLMs) to agentic systems. We outline the working of LLMs and agentic systems and present their differences. Recent prominent successes of AI systems have mainly been due to massive agentic systems that need to store information, plan, and coordinate and often use Wikis for that purpose. The lecture focuses on our recent agentic system for answering grammatical questions that builds empirically-supported answers based on Universal Dependencies language database for over 180 languages.

About the lecturer

Marko Robnik-Šikonja is a Professor of Computer Science and Informatics at the University of Ljubljana, Faculty of Computer and Information Science, and head of the Machine Learning and Language Technology Lab. His research interests span artificial intelligence, machine learning, natural language processing, and explainable artificial intelligence. He is a co-author of more than 250 scientific publications with over 11,000 citations. He has coordinated several national and EU research projects and developed many open data mining software packages and language resources.