What question did this study set out to answer?

The research aims to categorize and validate three types of world models in autonomous systems, focusing on the importance of auditability.

August 1, 2026Open Access

Three Forms of World Models: Perceptual-Explicit, Relational-Implicit, and Operational-Systemic — Framework and Evidence from an Autonomous Production System

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial explores autonomous system performance with three world model types, suggesting auditability precedes accuracy.

The research aims to categorize and validate three types of world models in autonomous systems, focusing on the importance of auditability.
Conducted three experiments over 72 hours on an autonomous production system with 973 scripts and 95 scheduled jobs.
Performed peak-shaving simulations, autonomous fault localization, and dual-source locking experiments.
Developed an orthogonal coordinate system to classify world models based on target, representation, and verifiability.
Peak-shaving simulation predicted a reduction from 39 to 12 tasks per minute, validated by independent measurements.
Successful autonomous fault localization and repair trial documented full pipeline recovery.
Dual-source locking showed independent recomputation at 83.94 ≈ 83.9, enhancing evidence credibility.

Zengbao Yu (2026) studied this question.

View Full Paper

Ask AI

Bookmark