Guessing.

When we give information we don't know to be true -which is what the model does when it builds low confidence tokens on each other- we call it guessing. Why would this be any different?

Guessing has a very mild implication of thought or estimation from a model of understanding of the subject in question.

Hallucination, on the other hand, is evocative of a less reasonable/rational source.

That would actually be an educated guess or a guesstimate. A guess is as good as any.

Maybe strictly definitionally, but comparing the connotations of the two, I assume people don't actually truly randomly guess.

Isn't that anthropomorphic too? It's humans doing guessing, not objects.

Not sure as I’m not yet familiar enough with the types of errors occurring. I do think a somewhat more accurate and possibly technical set of terms would be more useful for discussions about it though than hallucination.

They're not errors.

These are token prediction models. Their token predictions are accurate, insofar as the probabilities of those tokens in those sequences being representative of the kind of thing they were trained on.

Is it plausible that a document exists that explains that George Washington was actually an amateur magician? How is a GPT model to know that that is not the document it is predicting the next tokens for? Why wouldn't it explain that he used to perform for children's parties, and do tricks involving making doves and rabbits appear?

Exactly right. These are statistical models that operate at a completely different level of abstraction from the human brain: tokens/words/etc for text models, and individual (or perhaps small groups) of pixels for image models. Plausible output is obtained through sheer brute-force, and there is no mechanism to correct what we perceive as errors.

confabulation

Tripping

Bugs, faults, errors

Bullshitting

Or "high-loss example"

Bullshit implies active intent to deceive about your level of knowledge/understanding. In other words, it's even more anthropomorphic.

Yes I agree now. "High loss example" while not catchy might be closer. Although it is a bit backwards as you only think about creating the target once you sense the answer is not great.

Mistakes