Excellence in Research: Towards Responsible and Dependable Neuro-Symbolic Artificial General Intelligence for Human-AI Teaming

This project seeks to develop a trustworthy and reliable hybrid artificial intelligence (AI) framework that integrates the adaptive learning capabilities of advanced neural networks with the reasoning, transparency, and safety of symbolic AI to create robust, explainable, and dependable intelligent systems. The resulting “neuro-symbolic” technologies will enable more effective human-AI collaboration in mission-critical domains, including national security, disaster response, autonomous systems, and defense. The project will also expand research and educational opportunities for undergraduate and graduate students, helping to prepare a highly skilled AI workforce to meet the growing needs of government, academia, and industry. The research will develop novel hybrid neuro-symbolic AI methods for scene understanding in unstructured environments and Neuro-Symbolic Reinforcement Learning (NSRL) algorithms that integrate perception, learning, reasoning, and decision-making to enable adaptive Human-AI collaboration. The resulting AI systems/agents will continuously learn from experience while providing transparent, verifiable, and explainable decisions to improve operational effectiveness, resilience, and user trust. The framework will be validated using existing and planned AI research infrastructure and representative mission-critical application scenarios to demonstrate scalability, robustness, dependability, and operational efficiency. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.