I think part of my deep-rooted distrust of AI (more accurately, LLMs) in the context of generating text has to do with provenance. We mostly think of provenance in the contest of objects, like antiquities or paintings or manuscripts. The provenance of an object includes who made it, and where and when, and who bought and sold and owned it (s. v. AHD provenance).
As a specialist in early literatures, I have a particular interest in the provenance of a text. Who created the text, who copied the text to a manuscript, when was the manuscript made or the text published, by whom, and where?
In formal scholarship, published or not, the citation provides the provenance of a reference, the supporting authority for an assertion or quotation or fact.
LLMs muddy the history of a text. LLMs don’t create. They regurgitate textual patterns. LLMs use textual frequency to generate predictable patterned text. We don’t know who created an LLM text, how many texts are uncited palimpsests of that text, how many other writers’ words are being plagiarized? How many other writers will have identical sentences ?
I do not use generative AI or LLMs for my writing or images or research.
I wrote every word and created every image on my sites, except for those words or images by other people, whom I cite and or credit.
I do not use AI to read or summarize or generate emails. I do not read any email I suspect was AI generated.
Pay someone to edit if you truly feel you can’t write well enough.
I do not consent or permit my sites or my content to be scraped by AIs or LLMs or their associated Web crawlers or bots. for training purposes, or cached or stored in databases for the training or use of AIs or LLMs or their toolsets, crawlers, and bots.
I actively block such LLM or AI training and caching bots and crawlers when possible.
August 11, 2026
Copyright 2024–2026 by Lisa L. Spangenberg