fpf.tools
Library / First Principles Framework (FPF) - Core Conceptual Specification
Jump to passage
In this reading
Source · lines 61916–61926

Link to current text

Published source confirmed at last check

Source changed 2026-10-03 08:25:59 UTC · snapshot created 2026-10-03 08:26:43 UTC · last check 2026-10-03 09:40:10 UTC

C.26:8.2 - AI and LLM work-cycle route examples

LLM-mediated work cycles often create the same representational mistakes C.26 repairs: false passive read, false faithful summary, false shared comparison frame, and shortcut without loss/use declaration.

AI caseRoute
LLM summary of an architecture recordA.6.3.CSC, A.6.3.RT, and A.10 first; C.26 coarsening only if a state-representation shortcut is being overused.
Prompted model evaluation changes model or prompt behaviorC.16 / B.3 first; C.26.1 only if the eval output is treated as a passive model-state read.
Agent work cycle “discovers” requirementsA.15 / A.10 / A.6 first; C.26.1 only if the interaction created the requirement framing.
Synthetic personas “represent market state”A.10 / B.3 first; C.26.2 only if a low-recoverability state-reading claim is carefully bounded with non-admissible use visible.