A new study found that artificial intelligence responds most accurately to Polish commands, leaving English far behind in sixth place. Researchers from the University of Maryland and Microsoft compared how AI systems handled 26 different languages to determine which produced the clearest responses.
The researchers described their results as “surprising and counterintuitive.” They noted that English, often assumed to be the best-performing language for AI, failed to top any category. Polish consistently achieved the highest comprehension levels across every tested model.
The team evaluated leading AI systems such as OpenAI, Google Gemini, Qwen, Llama, and DeepSeek, feeding them identical inputs in each language to measure precision and consistency.
Machines Master a Language Humans Find Difficult
Polish achieved an impressive 88% accuracy rate when completing assigned tasks, the highest among all languages tested. The Polish Patent Office commented online, “AI follows Polish commands with unmatched precision. Humans find the language complex, yet AI handles it easily.”
This outcome challenges assumptions about data availability. Although AI models train on far less Polish text than English or Chinese, they still demonstrate strong comprehension. Meanwhile, Chinese placed near the bottom, finishing fourth from last in the ranking.
The Ten Languages AI Understands Best
The research ranked the top-performing languages for conversational clarity:
- Polish — 88%
- French — 87%
- Italian — 86%
- Spanish — 85%
- Russian — 84%
- English — 83.9%
- Ukrainian — 83.5%
- Portuguese — 82%
- German — 81%
- Dutch — 80%
Researchers concluded that linguistic complexity does not dictate AI understanding. The results suggest that machines interpret Polish—a language long known for its difficulty—more effectively than many humans can.
