SIGNAL

Philosopher Erik Larson argues that decades of philosophical influence have misdirected AI development, particularly through ontological approaches that prioritize formal knowledge representation over practical reasoning. The core issue centers on how philosophers shaped AI systems to rely on explicit, categorical knowledge structures rather than the more fluid, probabilistic reasoning patterns that appear more aligned with human cognition. This misalignment may explain persistent gaps between AI capabilities and human-like understanding, affecting everything from language comprehension to common-sense reasoning. Larson's self-critical analysis suggests the AI field may need to reconsider foundational assumptions inherited from philosophy and reorient toward empirical, data-driven paradigms.

Generative AI in educational settings risks cognitive atrophy by removing productive struggle from the learning process, analogous to how machines eliminate physical exertion. Furze proposes a "resistance framework" built on five mechanisms—expertise validation, critical evaluation, metacognitive awareness, deliberate cognitive stretch, and structured feedback—to preserve learning integrity while leveraging AI tools. The approach treats resistance not as friction to minimize but as essential scaffolding that maintains the cognitive load necessary for skill acquisition and deep understanding. This framework matters operationally because organizations deploying AI for training and development must architect it to maintain employee capability growth rather than enable performance shortcuts.