The Language of AI Could Change How Humans Speak

This essay argues that large language models, trained largely on written and scripted sources, risk reshaping how people speak and think by amplifying stylized, narrower, and sometimes inhuman language patterns; it warns of feedback loops as AI-generated text becomes training data, and highlights social harms including constrained vocabulary, increased curt or bossy speech, reinforced biases, and erosion of informal, authentic spoken communication.