MLST: Training AI to Replicate Papers — Faraday Beats Codex and Claude on Held-Out Tasks

Machine Learning Street Talk · rss · 2026-09-12

Machine Learning Street Talk hosts Edward Hughes, Chief Scientist and co-founder of Inherent, on whether machines can learn the judgment that separates plausible-looking results from faithful experiments.

Core arguments

The paper

Further topics include from-replication-to-innovation (how the Transformer happened), whether the AI scientist can Goodhart the judge, 8xB300 training runs, the RL crisis of per-turn credit in GRPO, recursive companies crossing a phase transition, and what replaces OKRs. Full timestamps and references available in the episode.

Original post →

More from Research

Research channel →