← Back to PeopleSBResearcherSteven ByrnesArtificial General Intelligence safety researcher at LessWrongWebsiteSteven Byrnes is a researcher focused on artificial general intelligence safety research.1Mentions1Articles1OutletsArticlesMost recent firstMentioned InReferenced in coveragePretraining data, not verifiability, is why LLMs are especially good at math (and coding)LessWrongneutralSep 21, 2026“Author of analysis on why LLMs are good at math, arguing pretraining data quality matters more than verifiability.”