I.J. Good

British mathematician and statistician (1916–2009) who worked with Alan Turing breaking Enigma at Bletchley Park, where the two developed what became known as Good-Turing frequency estimation — a method for estimating the probability mass of unseen events from how often items in a sample occur exactly once. He later coined "intelligence explosion" (1965) in his writing on an "ultraintelligent machine," and in 1952 introduced the logarithmic scoring rule now used, under a different name, as the cross-entropy loss that trains language models. Matters to this vault as a recurring foundational figure across three independent threads that keep converging on him from different directions: the statistics of the unseen, the theoretical roots of AI-risk vocabulary, and the mechanics of how LLMs are actually trained. The Good-Turing missing-mass estimate is also the explicit statistical core of Kalai & Vempala's 2023/2024 hallucination-rate theorem — a fourth independent thread reaching for the same estimator.

References

• claim-ij-good-1998-reversed-intelligence-explosion-to-extinction
• claim-ij-good-1965-ultraintelligent-machine-coined-intelligence-explosion
• claim-ij-good-1952-logarithmic-score-is-the-llm-cross-entropy-loss
• claim-unseen-species-problem-has-dual-origin-ecology-and-cryptanalysis
• claim-singletons-are-the-diagnostic-of-the-unseen
• claim-kalai-vempala-2023-hallucination-bound-is-a-rate-theorem-sorted-by-fact-regularity
• Captures: 10-inbox/raw/2026-07-11-hop-ij-good-loss-and-intelligence-explosion.md, 10-inbox/raw/2026-07-11-hop-estimating-the-unseen.md, 10-inbox/raw/2026-11-02-this-weeks-ai-safety-capture-found-llm-errors.md