Toward Intelligent Sensing Systems: Non-Equilibrium Materials as Platforms for AI-Enabled Autonomous Discovery

Conventional sensing systems rely on sequential architectures in which signal acquisition, processing, and decision-making are physically and functionally separated. This paradigm imposes limitations in latency, energy efficiency, and adaptability, particularly in data-intensive and dynamic environments. In this perspective, we discuss an emerging framework for intelligent sensing systems in which these functions are increasingly integrated through the intrinsic properties of functional materials. Non-equilibrium materials exhibit nonlinearity, memory, temporal dynamics, and adaptive responses that enable in-sensor information transformation. When coupled with artificial intelligence, these material capabilities support sensing platforms capable of encoding, processing, and interpreting information at or near the point of measurement. We examine key material platforms, architectural strategies, and opportunities for closed-loop autonomous discovery, while also highlighting challenges related to variability, scalability, and system integration. This convergence of materials science and intelligent systems points toward sensing technologies that move beyond passive measurement toward adaptive, low-latency, and energy-efficient operation.