Why Don’t You Know? Evaluating the Impact of Uncertainty Sources on Uncertainty Quantification in LLMs
Maiya Goloburda, Roman Vashurin, Fedor Chernogorsky, Nurkhan Laiyk, Daniil Orel, Preslav Nakov, Maxim Panov · 2026
Read Why Don’t You Know? Evaluating the Impact of Uncertainty Sources on Uncertainty Quantification in LLMs by Maiya Goloburda, Roman Vashurin, Fedor Chernogorsky, Nurkhan Laiyk, Daniil Orel, Preslav Nakov, Maxim Panov (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-10495.txt.
Description
Goloburda and six colleagues split language model doubt by its source. Most scores blend all doubt into one number, hiding what kind of trouble lurks. Their new data set tags each case by source: knowledge gaps, output wobble, or input vagueness. Methods that shine on knowledge gaps alone stumble or mislead on the other kinds. The lesson is plain: honest uncertainty tools must name the source, not just the size.
Catalog record
| CREATOR | Maiya Goloburda, Roman Vashurin, Fedor Chernogorsky, Nurkhan Laiyk, Daniil Orel, Preslav Nakov, Maxim Panov |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 65925 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.10495