A groundbreaking study revealed that artificial intelligence interprets Polish more effectively than any other language, while English ranked only sixth. Conducted by researchers from the University of Maryland and Microsoft, the experiment compared 26 languages to identify which ones AI systems process with the greatest precision.
The results challenged assumptions about English dominance in AI. “English didn’t come first,” the researchers explained. “It placed sixth in long-text performance, while Polish consistently produced the most accurate results.” The team tested leading AI models from OpenAI, Google Gemini, Qwen, Llama, and DeepSeek under identical prompts translated across all languages.
AI Excels in One of the World’s Hardest Languages
Polish achieved an average task accuracy of 88%, the highest of all tested languages. The Polish Patent Office noted that AI follows Polish instructions with exceptional clarity, despite the language’s complexity for humans. “Polish turns out to be most precise for AI,” the office wrote. “People may struggle with it, but machines seem to thrive.”
Researchers highlighted the irony that AI mastered Polish despite having far less training data than English or Chinese. In contrast, Chinese performed near the bottom, finishing fourth from last out of all 26 languages.
The Languages AI Understands Most Clearly
The top ten languages for AI comprehension, listed by effectiveness, were Polish in first place, followed closely by French, Italian, and Spanish. Russian ranked fifth, while English came sixth, slightly ahead of Ukrainian and Portuguese. German and Dutch completed the top ten.
The researchers concluded that language difficulty does not always determine machine understanding. In a twist of irony, AI now interprets one of humanity’s hardest languages better than humans themselves.
