I'd be quite interested to see a study of some kind about this - for example, tracking diversity of word choice, phrasing patterns, rhetorical devices etc among human-authored text - compared over time among groups of heavy users / moderate users / non-users of LLMs.
Obviously the concept of "individuality" is fairly tricky to operationalise but still, it might be able to indicate a measurable effect and if so, that would be fascinating.
Obviously the concept of "individuality" is fairly tricky to operationalise but still, it might be able to indicate a measurable effect and if so, that would be fascinating.