Source: Shae O.
As AI-generated content saturates digital spaces, the linguistic patterns of large language models are bleeding into how humans actually write and speak—flattening idiom, narrowing vocabulary, and erasing regional and generational voice. Speech is identity. A monoculture of AI-optimized language strips away the signaling markers consumers use to find communities and express taste. The damage falls hardest on women and marginalized groups who historically use linguistic variation as a tool for belonging and resistance. The private model argument here is less about privacy and more about preventing a linguistic commons from collapsing into a single algorithmic dialect that serves only the companies running the largest models.