Google launches a pilot of double-blind AI evaluations, keeping external evaluations in a cryptographic "box" to stop benchmark contamination and protect IP (Google DeepMind)
Google DeepMind piloted cryptographic double-blind evaluations to reduce benchmark contamination while protecting external evaluators' private test data.
Excerpt
Google DeepMind:
Google launches a pilot of double-blind AI evaluations, keeping external evaluations in a cryptographic “box” to stop benchmark contamination and protect IP — Building trust in proprietary model benchmarks using cryptographically secure environments — Imagine a student is set to take a high-stakes exam.
Read at source: https://www.techmeme.com/260827/p36#a260827p36