Cross-linguistic patterns of cognitive biases in large language models: a comparative study in English, Hebrew, and Russian

Katya Grander, Victoria Mosub, Sigal Kordova · 2026

Read Cross-linguistic patterns of cognitive biases in large language models: a comparative study in English, Hebrew, and Russian by Katya Grander, Victoria Mosub, Sigal Kordova (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/PMC13318900.txt.

Description

Katya Grander, Victoria Mosub, and Sigal Kordova present a 2026 comparative study of thinking biases in large language models across languages. Three popular models, ChatGPT, Claude, and Gemini, tackled tasks probing availability and confirmation biases. The same tasks ran in English, Hebrew, and Russian. Human answers provided the baseline.

The dataset held 2,028 observations, split between 507 human replies and 1,521 model replies. Models beat humans on the rule-based deduction task. Yet they copied human-style errors on heuristic tasks. Effects shifted significantly by language and by model. That breaks any assumption of even multilingual performance. The authors urge cross-language testing before such systems support real decisions, especially where biased reasoning can distort judgement quality.

Catalog record

CREATORKatya Grander, Victoria Mosub, Sigal Kordova
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE97714
LICENSEhttps://creativecommons.org/licenses/by/4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/article/MED/PMC13318900

Browse the full library · About the Tower of Babel