In the beginning of searchable data bases like Wikipedia and Google the problem was how to enter all that information. The fear was that we’d never get that huge backlog of information all loaded in and an “ignorant” incomplete library would lead to bias or manipulation. Now, these systems have evolved into AI Large Language Databases. As millions of people have queried the data bases, the questions and the answers have become part of the data that the models have learned to search, reword, and reply to.
The AI Large Language Databases are now copying words, phrases, sentences, paragraphs, and whole explanations of concepts. The bases are using computer speed to generate these things based on how often they are found.... IN THE DATA BASE!
They are now quoting themselves, and they are indefatigable enterers of data. They are using queries by billions of human questioners to answer things they have "seen" in the data base that they are rapidly becoming the majority contributor to. They are quoting their quotes of the quotes of themselves.
Look up "positive feed-back" and see what you find.... but the AI Large Language Databases may have already distorted the answer by running it through their tautological echo chamber of feed-back.
Gee, what would happen if you asked it whether or not it is quoting itself?
