Google DeepMind's Henry Shevlin argues LLMs are shifting human writing, with users actively avoiding or adopting machine-associated vocabulary · Digg