Autonomous AI agents, empowered by recent advances in Large Language Models (LLMs) and generative AI, are rapidly changing how we interact with, structure, and reason about information. These "agentic" systems go beyond static analysis – they act: interpreting their environment, interacting with humans and other agents, and autonomously executing tasks. This evolution opens new doors for knowledge engineering, especially in the context of increasingly complex, dynamic, and human-centric information spaces. While traditional knowledge engineering emphasized structured symbolic representations (e.g., ontologies, knowledge graphs) and formal reasoning, modern generative agents leverage machine-learned representations and natural language interfaces to emulate intelligent behavior. However, this comes with new challenges: How can we ensure such agents are interpretable, controllable, and aligned with human expectations? What is the role of hybrid systems that combine symbolic reasoning and statistical learning? This meeting will explore agentic AI as a foundation for the next generation of knowledge engineering systems – those that not only understand and transform data, but can also autonomously interact, reason, and collaborate.

Goals of the meeting:

• Understanding the potential and limitations of agentic AI systems – particularly LLM-based agents – for knowledge engineering tasks.
• Exploring hybrid approaches that combine symbolic and sub-symbolic methods to enable interpretable, controllable, and robust agent behavior.
• Identifying research needs, applications, and best practices for deploying agentic AI systems that align with human values and domain-specific requirements.

Creative Commons BY 4.0

Achim Rettinger