A key finding of the study is that the decline of linguistic diversity started well before European colonialism. The "golden age" ended between roughly 3,000 and 1,000 years ago, driven primarily by the expansion of agriculture and the rise of larger, more centralized populations that absorbed or replaced smaller language groups .
The researchers also estimated that at the beginning of the Holocene (~12,000 years ago), only about 4,500 to 6,200 languages existed — fewer than today — because human populations were much smaller . This contradicts the idea that linguistic diversity has steadily declined since the last Ice Age; instead, it grew alongside population before peaking and then falling.
Today roughly 7,500 languages are spoken or signed globally . But the losses are accelerating. Complementary research published in Nature Ecology & Evolution (not the Science paper itself) projects that without intervention, language loss could triple within 40 years, with at least one language lost per month and over 1,500 languages potentially lost by the end of the century
.
The Science study focuses on historical reconstruction and does not directly model AI's impact. However, the wider research context makes clear that artificial intelligence is exacerbating the marginalization of smaller languages.
Research from Johns Hopkins University presented at the 2025 Conference of the Association for Computational Linguistics found that popular LLMs are building "information cocoons" rather than leveling the playing field . A Nature article from November 2025 notes that languages spoken by communities in Asia, Africa and the Americas — accounting for about 5,500 of the world's ~7,000 languages — are "at risk of disappearing by the end of the century"
.
The Science study reshapes our understanding of human linguistic history. Diversity was once vastly greater than today, peaking at tens of thousands of languages around 2,000 years ago. The decline began not with colonialism but with agriculture. And while the historical forces are millennia old, modern AI systems risk accelerating the loss by excluding most of the world's languages from the digital future.