kalai-ideal-data-assumption
IN premise — summaries/2026/08/24/kalai-2023-hallucination-inevitable-s0-abstract.md
Created 2026-08-24T17:10:58+00:00
The hallucination lower bound for singleton facts holds even assuming ideal training data containing no errors whatsoever, ruling out data-quality explanations
Summary
Even if the training data were completely flawless with zero mistakes, AI models would still hit a mathematical floor on how often they generate false single-fact claims. This rules out "just clean up the data" as a fix and points to a deeper structural limitation in how these models represent isolated facts.