Keynote Speakers

The growing presence of intelligent systems in domains such as healthcare, mobility, agriculture, energy, and public infrastructure has changed the nature of engineering. The challenge is no longer to build isolated functionalities, but to shape complex socio-technical ecosystems. Although advances in artificial intelligence have greatly improved perception, prediction, and optimization, many deployed "smart" solutions still behave unintelligently at the system level. The problem is often not a lack of intelligence in individual components, but a lack of systems thinking in their engineering. This keynote examines current systems engineering challenges from a systems thinking perspective and discusses the integration of AI with classical systems engineering principles. Rather than treating AI as a module added to an existing structure, intelligence is viewed as a property that emerges from the coordinated behavior of the whole. Through illustrative cases, the talk highlights recurring failure patterns rooted in reductionistic development and shows how systemic design patterns can address them. The keynote concludes with implications for research and education: progress in intelligent systems depends less on training better models and more on designing better systems.

Artificial intelligence is increasingly embedded in everyday environments populated by sensors, connected objects, and services. Applications range from smart homes and assistive technologies to industrial systems and social robots. While these intelligent ecosystems promise efficiency and personalisation, they often remain opaque, difficult to control, and poorly aligned with human values and everyday practices. This talk introduces the concept of humanations: human-understandable, controllable automations that enable people to shape intelligent behaviour rather than merely adapt to it. Building on research in human-computer interaction, end-user development, explainable AI, and intelligent environments, the talk presents design concepts and prototypes that foster transparency, intelligibility, and meaningful user control in real-world settings.

Artificial Intelligence (AI) and the Internet of Things (IoT) have long been recognised as foundational technologies shaping future digital society. Their convergence, often referred to as Artificial Intelligence powered Internet of Things (AIoT), has accelerated the deployment of intelligent, data-driven services across cyber-physical environments. This talk examines the structural evolution of the Internet toward an Agentic AI powered Decentralized Internet, moving beyond technology-centric integration toward a paradigm-level transformation of digital ecosystems, exploring how autonomous AI agents become first-class entities that perceive, reason, coordinate, and act across distributed environments, and outlining future research directions toward a trustworthy, human-centric, and value-oriented Agentic AI powered Decentralized Internet.

Trustworthy autonomy ultimately hinges on safe perception: autonomous decisions are only as reliable as the sensing and inference pipelines that support them, especially under disturbances, faults, and attacks, while meeting quantitative risk constraints typical of safety-critical domains. In this talk, trustworthy autonomy is framed as justifiable autonomy. The core message is that Model-Based Engineering and Digital Twins can provide a rigorous foundation to engineer and assure safe perception, presented as predictive run-time models enabling continuous monitoring, planning, and safe reconfiguration, integrated into an autonomic MAPE-K loop and organized hierarchically across multiple system levels.