In today's column, I explore concerns about the potential influence of generative AI and large language models (LLMs) on human vocabulary. Some worry that AI will prioritize certain words over others, shifting our vocabulary toward those favored by AI while neglecting others. Additionally, AI could introduce new, seemingly made-up words into common use, which people might accept as real due to repeated exposure. This transition could lead to an AI-devised vocabulary over time. Words are essential for communication, yet they evolve. Societies elevate some while others fade. For example, many Shakespearean words are no longer in use. Our lexicon grows from about 50 words by age one to around 40, 000 in adulthood, yet newspapers use only about 20, 000 unique words, shaping our exposure and usage of language. As generative AI becomes more prevalent, it's critical to consider the words it chooses.

AI systems select from limited data sets, potentially omitting rare or untrained words. This selection subtly influences human vocabulary, presenting a vocabulary users might unknowingly adopt. Over time, this could homogenize language, simplify vocabulary, and introduce AI-coined terms. To mitigate unexpected impacts, awareness of the linguistic changes AI may cause is crucial. Experts and policymakers can address these subtle shifts and the introduction of AI-generated words into human vernacular. Research on human word evolution highlights constant linguistic change driven by communication efficiency. AI could accelerate this evolution, providing a streamlined language for an interconnected world. It might lead to a more unified, simplified means of communication, potentially enhancing linguistic innovation. Despite fears of AI-altered vocabulary, this evolution could have positive outcomes, perhaps ushering in a new era of efficient and kind language. Mother Teresa noted that even short, kind words leave lasting echoes; AI catalyzing such linguistic shifts may amplify this impact.