In Search of Missing Mass: Disentangling Discoveries from Hallucinations @SimonsInstitute
In Search of Missing Mass: Disentangling Discoveries from Hallucinations  @SimonsInstitute
Uploaded October 2026 | Updated November 2026, 4 days ago
Eric Nalisnick (Johns Hopkins University)
https://simons.berkeley.edu/talks/eric-nalisnick-johns-hopkins-university-2026-10-07
Trustworthy AI: From Hallucinations to Reliable Autonomy

Large language models (LLMs) hold much promise for scientific discovery. However, many basic questions remain about when an LLM can reliably make discoveries. In this talk, I will discuss how to interpret and extend the hallucination results of Kalai & Vempala [STOC 2024] to characterize an LLM's potential for discovery. In particular, how might we identify prompts that are likely to yield discoveries rather than hallucinations?
In Search of Missing Mass: Disentangling Discoveries from HallucinationsExplicit lossless vertex expandersQuantum ergodicity on graphs (Part 2)Convex-body Sampling to the Kannan-Lovász-Simonovits ConjectureBuilding a Quantum Computer with QLDPC CodesSampling elements of a finite group: efficiency of the product replacement algorithm with...The power of subspace designs: Applications in coding theory, pseudorandomness, and beyondLightning TalksDiffusion in RL and robotics: how expressive policies changed how we use continuous actionsField dynamics and its applicationsOutlier singular values and (de)localized singular vectors of sparse random matricesUnderstanding Language Model Hallucinations through Parametric Knowledge Integration
Simons Institute for the Theory of Computing |

In Search of Missing Mass: Disentangling Discoveries from Hallucinations

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER