2000+ years ago philosophers like Socrates and Epictetus told their followers not to write things down. They believed that by writing things down we did not fully understand things as we didn't think carefully about what we wrote.
Then lots of people wrote lots of things down anyway. Its quite likely that they did not have a complete knowledge of what they were thinking and writing about but what was written down was accepted as correct and true. Other people read the less than correct writing and "learned" from it.
The movable type came along and vast amounts of knowledge was turned into books. Again what was written down might not be what was intended desired People were stuck with trying to learn from texts written by people who might well not understand what they were writing about
Next came the internet. People who definitely didn't understand what they were talking about could write anything they damn well like. Other people could read this and there in no easy way for even an expert to tell if the text is factually accurate or even consistent at times. (cf. Post Modernist writers who definitely didn't help in this regard)
Now we have "AI" which doesn't understand the meanings of the words it spews across the globe. It only understands that those individual text strings are often associated together in text stolen from all of the world's intellectual property. It "hallucinates" complete garbage and that is apparently a feature not a bug. It makes statements that are not only provably wrong but which are incredibly dangerous if someone believed them and acted on them.
So in two thousand years we have gone from "Don't write anything down as it will mean you have incomplete knowledge of the subject you are thinking about" to "This might well be just gibberish or incredibly wrong in highly dangerous ways. Who cares, you don't need to think any more anyway. Now pay us all your money for crap you don't want or need and doesn't work properly anyway."
Model collapse can't come soon enough. Of course it might have happened already, how would we know after all?
#AI #Knowledge #truth #LLM #GarbageInGarbageOut