Educating for Trustworthy Hybrid AI: Curriculum-Level Hybridization in an MSc AI Program

Abstract

The rapid development of Large Language Models (LLMs), Knowledge Graphs (KGs), and other heterogeneous AI technologies is changing not only how intelligent systems are designed but also how AI should be taught. As trustworthy AI increasingly emerges from the interaction among neural, generative, symbolic, and agentic paradigms, AI education faces a corresponding challenge: students must learn individual AI approaches and how to deliberately combine them into responsible hybrid intelligent systems. This paper develops a conceptual framework for hybrid AI education and illustrates its curriculum-level realization through the International Master’s Program in AI at the University of Jyväskylä. Building on three educational challenges identified through the KGELL COST Action’s WG5 work on ethical, privacy, and bias (EPB) issues, the study addresses the need to hybridize AI education in response to hybrid AI architectures, the distinction between responsible AI design and responsible AI use, and the continuing educational role of KGs, ontologies, and symbolic AI in the era of increasingly capable LLMs. The case study showed how these principles can be embedded across a curriculum organized around complementary AI paradigms, supported by cross-cutting competencies in safe, transparent, and responsible AI through the Cognitive Service Development Project. Observations from student-developed services further illustrated how heterogeneous AI paradigms can be combined in practice. The study, therefore, contributes a curriculum-level hybridization framework and an illustrative educational case showing how AI education can move from teaching heterogeneous paradigms separately toward developing the competence to integrate them into responsible hybrid human-AI ecosystems.