← back
arXivMartin Böckling, Elizaveta Nosova, Heiko Paulheim, Andreea IanaTue, Aug 4, 2026, 9:18 AM PDT
score 17.0

New multilingual test shows AI chatbots struggle with geography math

Original: MultiGlobeQA: A Multilingual and Globally Diverse Benchmark for Geospatial Reasoning

Source: arxiv.org

Writing ELI5 summary…