Author of 'From Prehistory to the Age of AGI' Kai Taraporevala Proposes 'Exocognitive Stage': AI Bypasses Thought, Results Effectively Undiscovered
Kai Taraporevala, a writer trained as a physicist who spent decades in corporate finance and investing, published a long essay on LessWrong borrowing neuroanthropologist Merlin Donald's four-stage framework of cognitive development to argue that AI marks a fifth 'exocognitive stage': unlike books and other external tools that support human thought while leaving humans as the drivers of cognition, AI directly bypasses the thinking process, making humans either co-creators or passive recipients. He contends that if AI solves a centuries-old problem using pattern-matching humans cannot comprehend, the result remains effectively 'undiscovered' for humans — unless a symbolic AI bridge is built to translate AI's intuition into traceable causal chains.
ImportanceMaterialEvidenceE3 inspectableWrite-upQuick
Neuroanthropologist Merlin Donald divides human cognitive development into four stages: episodic (learning from actions in the present), mimetic (copying others using language as explanation), mythic (with writing), and theoretic (further symbolic thinking and the empirical sciences). Kai Taraporevala, author of From Prehistory to the Age of AGI, argues that AI marks a fifth stage — the 'exocognitive stage.' The distinction is precise: in the theoretic stage, external tools like books support us but leave us as the drivers of cognition; in the exocognitive stage, AIs bypass our thoughts, and we become either co-creators or passive recipients of another intelligence's output.
From this he draws an epistemic consequence. If an AI solves a centuries-old problem using advanced multi-dimensional pattern-matching shorthand that we cannot comprehend, the result remains effectively undiscovered for us. Without a symbolic bridge, the mathematician-philosopher Jacob Bronowski's 'moment of appreciation' is broken — if we cannot re-create the theorem, we will no longer appreciate the knowledge. Taraporevala proposes that mathematicians develop symbolic AI to translate LLM proofs and alien 'intuition' into concepts and causal links that can be followed by humans.