Emre Barut

Posts 6

• October 2026A Statistical Analysis of LLM Benchmaxxing

  Do models score better on benchmarks that existed before they did? An estimate, lab by lab.

• December 2025Token Laundering

  Training on GPT outputs by way of an MIT-licensed model like DeepSeek, and why it works.

• June 2025My AI Stack (June 2025)

  My monthly AI tool stack for June: no changes from May.

• June 2025Statistics in the Age of AI: Reflections from the GW Panel

  My answers on the GW Statistics 90th anniversary panel on statistics in the age of AI.

• May 2025My AI Stack (May 2025)

  The first of a series: the AI tools I pay for and use, and why.

• April 2025How Accurate Are Lords of Limited's Early Format Guesses? An AI Analysis

  I had AI score Lords of Limited's early format takes against their own retrospectives.