AI, Judgment, and Nonconceptual Content: A Critique of Dreyfus in Light of Neuro-Symbolic AI

This paper examines Hubert Dreyfus' phenomenological critique of AI in light of contemporary large language models (LLMs) and emerging hybrid neuro-symbolic systems. While Dreyfus championed connectionist approaches over rule-based AI (GOFAI), his nonconceptualist framework faces limitations when applied to modern hybrid systems that combine neural networks with symbolic reasoning. I argue that Dreyfus' failure to account for intentional content in absorbed coping creates problems for explaining how nonconceptual significance can constrain the conceptual—a crucial issue for evaluating hybrid AI systems that must rationally connect statistical learning with rule-based reasoning. Drawing on later Husserlian phenomenology, the paper proposes a framework that preserves nonconceptual content while explaining its connection to the conceptual through an extended conception of judgment and rationality. This hybrid phenomenological approach better addresses the challenges posed by AI systems that integrate both connectionist and symbolic architectures.