The structural problem that AI-augmented workflows remove both the tedium of preparatory work and the formative friction embedded within it — the debugging session that built system intuition, the manuscript hunt that taught disciplinary structure.
Licklider's 85/15 ratio assumed the 85% was cognitive overhead — waste that could be offloaded without loss. The actual 85% contains, unpredictably distributed, the friction moments that build expert intuition. The debugging session that teaches how two subsystems interact. The dependency conflict that forces comprehension of architectural load-bearing structures. The performance regression that reveals assumptions the engineer did not know she was making. These moments are rare — perhaps ten minutes within a four-hour block — and they cannot be separated in advance from the tedium surrounding them. When AI absorbs the 85%, it removes both. The engineer does not choose to abandon the formative friction; she chooses to abandon the tedium, and the friction comes along as unnoticed luggage.
The Developmental Paradox
In The You On AI Field Guide
The Trivandrum engineer who discovered her architectural intuition thinning months after the coupling removed implementation work is the canonical illustration. Her intuition had been constructed geologically — each tedious debugging session