ON AI AND SOME PHILOSOPHICAL CHALLENGES

Abstract

This paper consists of two interconnected parts, both challenging mainstream computational views of mind and intelligence. In the first part, the author argues for a fundamental reconceptualization of cognition. He contends that the success of neural network-based AI and the universal properties of human language expose key limitations of classical computational theories of mind. Instead of viewing intelligence as rule-based symbol manipulation, he proposes that cognitive abilities emerge from architecturally constrained networks that mediate between analog and digital representations. This framework reconciles connectionist and universal-grammar approaches, suggesting that learning, innateness, and representation are best understood through structural constraints and emergent dynamics. Intelligence—whether biological or artificial—should therefore be seen as a product of architecture and emergence, not of explicit computation. In the second part, the authors examine the functional role of consciousness in biological organisms. They observe that current AI systems lack consciousness, but argue that consciousness-like mechanisms, such as real-time processing, reflexive awareness, and integrative functions, may become essential for AI systems that need to survive and adapt in dynamic, threatening environments. While they outline potential pathways for implementing such features, they also stress that artificial analogues would remain fundamentally different from biological consciousness. 156 Taken together, the paper offers a unified vision: intelligence arises from constrained architectures and emergent patterns, and as AI systems become more autonomous, they may require surrogate forms of consciousness to handle complex, real-world challenges—though these will never fully replicate the biological original.