Polish has emerged as the top-performing language for completing advanced artificial intelligence tasks, outstripping global lingua francas including English and Chinese, according to a new international study. The findings, revealed by researchers from Microsoft, the University of Maryland, and the University of Massachusetts Amherst, challenge established assumptions about language dominance in the rapidly evolving world of large language models.
The study assessed how the largest AI language models—a core component of modern chatbots and virtual assistants—handle intricate language processing across 26 different tongues. To the researchers’ surprise, Polish ranked first in accuracy for highly complex tasks, while English, historically the primary language for AI training data, slid to sixth position. Chinese, another major world language, performed even less impressively.
“Our analysis showed that, contrary to expectations, Polish leads in language model performance during the most demanding tasks, with an accuracy rate of 88 percent,” said Dr. Yekyung Kim, one of the study’s lead authors. “English and Chinese are typically assumed to be ahead, given their prevalence in AI datasets, but the outcome was surprising even for us.”
At the heart of the research were large language models (LLMs)—AI systems trained on massive corpora spanning news articles, literature, and internet content. These models underpin tools such as ChatGPT, Gemini, and Llama. When tested across a spectrum of languages and tasked with processing increasingly lengthy and complex texts, Polish proved most adept at managing information, surpassing other major European and world languages.
Context and Methodology
The study pitted six industry-leading AI models against one another, feeding them parallel inputs in languages ranging from Spanish and French to Tamil and Swahili. While Russian, French, Italian, and Spanish rounded out the top five, the worst results were posted by languages from the Bantu family—Sethoto and Swahili—despite their vast speaker bases.
Researchers theorized that the structure of Polish and its linguistic relatives may offer unique advantages for machine processing. “It appears that script type and the abundance of training data matter, but the capacity of Slavic languages, especially Polish, to convey complex meaning may also be a factor worth further investigation,” noted Jenna Russell, a co-author based at the University of Maryland.
The performance gap also track with language families: Slavic, Romance, and Germanic languages saw the strongest results, all utilizing Latin or Cyrillic scripts. In contrast, Bantu languages, despite having over 350 million speakers collectively, faced hurdles, likely linked to being considered “low resource”—with less data available for effective AI training.
Poland’s Investments in AI
Poland has invested in domestic AI capabilities, most notably with the government’s launch earlier this year of the Polish Large Language Model (PLLuM). This freely available tool is designed to underpin public sector services, including writing official correspondence, analyzing citizen inquiries, and automating document summarization. The city of Częstochowa became the pioneer in deploying PLLuM technology to speed up municipal workflows.
“Our goal with the development of PLLuM is to ensure Polish language and culture are meaningfully present in the digital age,” said Marzena Karpińska, a project coordinator for Poland’s national AI strategy. “These research results validate the work we’ve put into making Polish AI-ready and adaptable to the world’s most sophisticated computational tools.”
Poland’s move is part of a wider trend in Europe to ensure that local languages are not overshadowed by English-centric AI. By investing in specific linguistic models and encouraging open access, the Polish government aims to foster both technological innovation and cultural preservation.
Broader Implications
The emergence of Polish as an AI frontrunner has broad repercussions for international businesses, academic researchers, and governments seeking to deploy AI tools in diverse linguistic settings. It also suggests that investing in language-specific datasets and tools might yield competitive advantages for nations beyond the English-speaking world.
“This isn’t just about national pride,” said Dr. Kim. “What our research reveals is that linguistic diversity has much to offer the global AI ecosystem. Languages previously considered peripheral could be at the center of next-generation digital solutions.”
With AI expected to play an ever-greater role in shaping everything from government to education and commerce, the Polish example may spur other countries to reconsider their own digital priorities.
The world of artificial intelligence is opening new frontiers for languages beyond English, underscoring the value of linguistic and cultural plurality in a digital era. The study’s findings suggest that, as AI becomes more embedded in everyday life, smaller language communities like Poland’s may not only keep up but, in some cases, lead the way.
