New research titled “We Politely Insist: Your LLM Must Learn the Persian Art of Taarof” reveals that mainstream AI language models struggle with the Persian cultural practice of taarof, navigating such situations correctly only 34 to 42 percent of the time, compared to 82 percent for native speakers.
- Taarof governs daily interactions in Persian culture.
- AI models navigate taarof situations only 34 to 42 percent.
- Study led by Nikta Gohari Sadr of Brock University.
- TAAROFBENCH measures AI systems' taarof reproduction.
- Cultural missteps can derail negotiations and damage relationships.
- Taarof involves ritualized exchanges of offer and refusal.
The study, led by Nikta Gohari Sadr of Brock University, introduces “TAAROFBENCH,” a benchmark for measuring AI systems’ ability to replicate taarof. The findings indicate that AI models default to Western-style directness, missing crucial cultural cues.
Cultural missteps in high-consequence settings can derail negotiations and damage relationships. The researchers emphasize that AI’s cultural blindness could represent a significant limitation in global contexts.
“Taarof, a core element of Persian etiquette, is a system of ritual politeness where what is said often differs from what is meant.”
— Researchers
Originally reported by arstechnica.com as “When “no” means “yes”: Why AI chatbots can’t process Persian social etiquette” on 2025-09-24 02:23:00.
