A recent study by the University of Maryland and Microsoft compared how various AI models respond to prompts in 26 languages, including GPT-5, Gemini, Qwen, Llama, and DeepSeek. Surprisingly, Polish outperformed all other languages with an impressive 88% accuracy rate for task completion, leaving even English in the sixth position.
The study revealed the top 10 languages based on accuracy, with Polish leading the list at 88%, followed by French, Italian, Spanish, and Russian. Conversely, English ranked sixth with 83.9% accuracy, highlighting the unexpected dominance of Polish in AI prompting efficiency. Additionally, the study found Chinese to be among the least effective languages for AI prompts, ranking fourth from the bottom. This research suggests that utilizing the Polish language could significantly enhance prompt quality in AI interactions.
