Robustifying generative AI through human-centric integration
of neural and symbolic methods

Generative Artificial Intelligence (GenAI), such as foundation models, represents a powerful and transformative class of AI capable of learning patterns from data and generating new content. However, GenAI has notable shortcomings including hallucinations or bias which can lead to misuse or hinder its widespread adoption and positive societal and economic impact. These shortcomings stem from three key areas: technical, operational, and user robustness.

RobustifAI focuses on foundation models used in the context of human cyber-physical systems (HCPS) which are complex systems that combine computation, networking, humans and physical processes to monitor and control real-world environments with applications in many sectors.

HCPS represent the most demanding systems to address the robustness of GenAI due to their immediate physical impact, criticality, real-time as well as human interaction requirements. By tackling GenAI robustness in HCPS, RobustifAI will propose solutions applicable across various domains, ultimately unlocking the full potential of GenAI in conformity with ethical considerations.