Proves platform-determinism is necessary for trustworthy AI and implements an integer-only engine for bitwise identical inference across ARM and x86.
March 27, 2026
Original Paper
On the Foundations of Trustworthy Artificial Intelligence
arXiv · 2603.24904
The Takeaway
By resolving the non-determinism inherent in IEEE 754 floating-point math, this work enables verifiable AI outputs that are identical regardless of hardware. This is a prerequisite for on-chain AI verification and robust safety auditing across distributed systems.
From the abstract
We prove that platform-deterministic inference is necessary and sufficient fortrustworthy AI. We formalize this as the Determinism Thesis and introduce trustentropy to quantify the cost of non-determinism, proving that verification failureprobability equals 1 - 2^{-H_T} exactly. We prove a Determinism-VerificationCollapse: verification under determinism requires O(1) hash comparison; without it,the verifier faces an intractable membership problem. IEEE 754 floating-pointarithmetic fundamentally