Cohere's CulturalRiddles Benchmark Exposes AI's Cultural Knowledge Gaps

Cohere Labs released CulturalRiddles, a benchmark of 6,200 community-written riddles in 51 languages created by 167 researchers from 45 countries. Testing 28 AI models revealed major cultural knowledge gaps, with the best model scoring only 79 and accuracy dropping to 22% for some communities.

2026-10-26 ~ 2026-10-28 · 2 related posts

• Cohere Labs' CulturalRiddles: top AI model hits 79%, some cultures just 22% — Cohere_Labs · 2026-10-26
• Cohere Labs' CulturalRiddles: 6,200 Riddles in 51 Languages Expose AI's Cultural Blind Spots — Cohere_Labs · 2026-10-28