AI Language Trends: Emphasizing The Same Words Across the Globe
AI is shaping language in ways we didn't expect. A study spanning 34 languages shows AI's influence beyond English, with certain verbs appearing almost universally.
AI isn't just changing how we work, it's changing how we speak. A recent study analyzed AI's impact on language across 34 languages, beyond the typical focus on English. The findings? AI has a linguistic thumbprint, and it’s pressing down hard.
Global Word Choices
The research revealed that AI doesn't just dabble in English. It leaves traces everywhere. By looking at the WMT News Crawl corpus, the study identified that certain verbs, like those related to 'emphasize,' pop up in 24 out of 34 languages. That's not a coincidence. It's a pattern. AI's lexical preferences are spreading, and they're not playing favorites.
Using log prevalence ratios, researchers ranked AI-overused lemmas. This means they pinpointed which words AI loves to repeat. The increase in these AI-preferred terms was staggering. From 2020-2021 to 2023-2024, their presence jumped by 15.1% in 26 out of the 34 languages examined. Meanwhile, baseline words actually declined by 4.5%. Clearly, AI is nudging languages in a specific direction.
A Post-ChatGPT World
ChatGPT's release marked a turning point. Before and after comparisons show a noticeable uptake in AI’s favorite words. In ten languages with longer historical data, post-2022 saw noticeable increases, dwarfing earlier changes. Yet, the shifts were still smaller than those seen in Scientific English. It seems AI's impact is far-reaching, but it's nuanced.
The real story here's about cross-lingual homogenization. Are we heading towards a world where AI drives linguistic uniformity? If so, what does that mean for cultural diversity in language? These aren't just academic questions. They touch on how we communicate in a globalized world.
Why Should We Care?
AI's influence on language isn't just a curiosity. It's a concern. Language is the backbone of culture. If AI starts dictating how we speak, we're looking at a future where linguistic diversity could diminish. The gap between the keynote and the cubicle is enormous, and so is the one between AI developers and everyday speakers. But as AI's lexicon becomes more prominent, that gap might start to close.
Here's what the internal Slack channel really looks like: people are noticing the change, and it's not always welcome. Language is personal, but AI's imprint is anything but.
Get AI news in your inbox
Daily digest of what matters in AI.