Promising or Dead End? How Two Epistemologies of Perception Determine the Trajectory of AI Abstract

This study investigates the ontological and epistemological relationship between artificial intelligence (AI) and human sensory perception, interrogating whether AI can theoretically surpass or achieve parity with human perceptual capacities at the foundational stage of cognition. Employing an analytical-comparative methodology, the research contrasts two distinct epistemological paradigms: Al-Farabi’s theory of the “Common Sense” (al-ḥiss al-mushtarik) and Maurice Merleau-Ponty’s phenomenology of the “Lived Body” (le corps vécu), with a focused analysis on how each framework accounts for the integration of the five sensory modalities. The findings reveal that Al-Farabi’s stage-based model of perception, which unifies discrete sensory inputs through a central cognitive faculty, structurally parallels the layered data-processing architectures of contemporary AI systems. Merleau-Ponty, however, rejects the notion of raw sensory data, framing perception as a pre-reflective, holistic, and embodied engagement with the world that resists algorithmic decomposition. Consequently, the study argues that the theoretical and practical trajectory of AI is epistemologically contingent: Al-Farabi’s model opens a viable horizon for AI development and potential perceptual convergence with humans, whereas Merleau-Ponty’s emphasis on bodily intentionality and lived experience establishes ontological boundaries for disembodied systems, ultimately constraining AI to functional simulation rather than genuine perceptual cognition.

Rather than merely describing two paradigms, this article argues that empirical evidence from developmental psychology and 4E cognition—specifically the “kitten experiment”—actively corroborates Merleau-Ponty’s embodied model, thereby shifting the burden of proof onto any computational paradigm that claims perceptual parity for AI.