Structural Certification for Reliable Physical Design with Language Models
2026-06-29 • Artificial Intelligence
Artificial IntelligenceMachine Learning
AI summaryⓘ
The authors show that even if a language model isn’t fully reliable, it can still help create trustworthy physical designs by separating suggestion and verification steps. They introduce a method called Physics-Anchored Certification (PHACT), where the model proposes designs, and a deterministic engine independently verifies them to ensure accuracy. Their approach prevents false confirmations by only accepting results derived strictly from fixed inputs. Tested under challenging conditions, their method never wrongly approved a flawed design.
language modelphysical designdeterministic enginePhysics-Anchored Certificationpropose-certify loopfault toleranceverificationadversarial trialscertification
Authors
Nakul Vyas, Iliya D. Stoev
Abstract
An unreliable language model can be made to produce reliable physical designs if the authority to assert is moved out of the model: the model proposes, and a deterministic engine alone certifies, returning certified, impossible, or unknown. We introduce Physics-Anchored Certification (PHACT), a propose-certify loop spanning five scientific domains, and identify what makes such a certificate trustworthy. A checker that accepts a model-supplied value can be forged; deriving the certified quantity from fixed inputs instead makes forgery impossible by construction. Across eighty adversarial trials spanning two models, two decoding temperatures, and a deliberately faulted engine, this contract produced zero false certifications.