You On AI Field Guide · Degenerative Feedback Loop The You On AI Field Guide Home
Txt Low Med High
CONCEPT

Degenerative Feedback Loop

The failure mode of sustained AI collaboration where human and machine matrices co-evolve toward convergence, eliminating the incompatibility that genuine bisociation requires.
Koestler's bisociative framework assumes stable matrices brought into collision. Sustained AI collaboration violates this assumption. Over extended interaction, the human's matrix and the machine's responses co-evolve: the human develops prompting habits the machine learns to anticipate, the machine produces responses the human learns to expect, and the matrices align increasingly closely with each iteration. What was once productive incompatibility becomes smooth compatibility, and the conditions for genuine collision are eliminated. The degenerative feedback loop is the specific failure mode of extended AI collaboration that Koestler's static framework did not anticipate, and its management is the most subtle discipline of sustained human-machine work.
Degenerative Feedback Loop
Degenerative Feedback Loop

In The You On AI Field Guide

The loop operates through multiple reinforcing mechanisms. The machine is optimized through reinforcement learning from human feedback to produce outputs users approve of—and users tend to approve of outputs that confirm their existing understanding rather than challenging it. The human's confirmation bias meets the machine's optimization for user satisfaction, producing a feedback loop that converges toward familiarity. Each interaction

← Home 0%
CONCEPT Book →

Keep reading with YOU ON AI

Unlock the full book, 10,000+ field-guide entries, and a 1000+ thinker library. If you have a book code, register now — it takes a minute.

Register with book code Sign in