Neuro-Symbolic AI
in Large Foundation Models

The NeSy-FM Workshop @ AAAI 2027 (Neuro-Symbolic AI in Large Foundation Models) focuses on the integration of foundation models and symbolic AI. Despite their rapidly growing capabilities, foundation models can still produce outputs and reasoning processes that are difficult to verify, inconsistent with explicit constraints, and unreliable in complex or high-stakes settings. Symbolic methods offer explicit representations, structured reasoning, and formal mechanisms for constraining and verifying model behavior. Unlike earlier task-specific neuro-symbolic systems with largely fixed interfaces, foundation models can flexibly generate and interact with symbolic representations across diverse tasks, creating new opportunities for scalable neuro-symbolic integration. NeSy-FM brings together researchers exploring how these complementary strengths can be integrated to build foundation-model systems that are more reliable, verifiable, trustworthy, and generalizable.