PERSON DIRECTORY
Max Bartolo
Max Bartolo appears in 1 indexed conversation across Machine Learning Street Talk. This directory brings every appearance, source, TL;DR, digest, and transcript into one searchable feed.
Reasoning, Robustness, and Human Feedback in AI - Max Bartolo (Cohere)
Bartolo’s robustness standard challenges benchmark-led reasoning claims: a model that fails once on a simple task may not be reasoning reliably.Influence-function evidence found factual answers concentrated around a few source documents, while reasoning drew distributed procedural influence, weakening retrieval-only explanations.Human feedback can reward assertiveness over factuality, making adversarial data, dynamic evaluation, calibration, and controllable test-time compute key monitoring priorities.