Post

This is the headline that sparked my request. I am a practicing lawyer and law professor. Lawyers are increasingly using AI to write legal briefs, but AI fabricates court cases, and lawyers cite them as true. AI presents false information at true; it fabricates. www.latimes.com/california/s...

What's confusing is that AI's appear to know some facts, like the height of LeBron James, a famous basketball player. But that's simply because text describing his height appears often in the training text. It sees 6 ft 9 near LeBron James often enough to repeat it -- but it doesn't "know" this fact

Thank you for that. That is indeed an insane basis for her claim. Here I thought it was something about respecting the rights of the schizophrenic or something. This is even sillier than that, though. “Please their task masters” my, my.

"AI scientists?"

We're calling it "intelligence." I think the comparison with humans ship has already sailed. "Hallucinating" evocatively describes the serious mistakes it makes.

Mental illness is hardly the only cause of hallucinations. And confabulation is present in several neurological and psychiatric disorders. So your substitution is a wash at best, if comparison with human illness is your standard.

mistakes

There is, in fact, a publishing industry standard term for statements which are orthogonal to all truth-vectors. "Bullshit." LLMs produce bullshit. There's no factual analysis possible or warranted of an LLM's output, because it wasn't assembled with regards to a truth structure.

"Fabricate" works better than "confabulate" for me

Glitch

AI functioning normally It can only incidentally create accurate information and any effort to disguise that fact is doing unearned pr for it

I'd call it inaccurate information.

I think it starts with popularizing using “A.i.” instead of Al. Then maybe we can give ”A.i.” an ”F.”

Delusion?

misreport

Maybe instead of a special term for when AI gets something wrong, we should have a special term when it gets something right I propose "lucky guess"

What is wrong with just saying that it was wrong or incorrect? Why does it need its own term?

LLMs neither have mental states nor the ability to evaluate truth, so any terms like "hallucinate" that imply they do are inappropriate. Some suggestions: Verb: to mangle, to garble, to fail, to error out, to gibber, to bullshit Noun: error, nonsense, gibberish, failure, bullshit

Fabricated works so well.

Slurrification

Hallucinations

I would have said "confabulation" except that (surprisingly, to me) the leading synonyms for that seem to be like "discussion". After a while I decided I liked "fabrication" best. I agree that "hallucination" seems to imply conscious experience.

Hallucinate. It’s already settled. I’ve never heard anyone use the other terms. Definitely hallucinate.

I think fabrication is great. All generative AI content is constructed (fabricated) independently of reality, since it's just based on what other people have said which may or may not be correct. And I think a fabrication (vs a lie) is a good way of describing that it's not corresponding to reality.

Garbage in, garbage out.

Errors. The code is supposedly designed to provide accurate answers. It’s an error when it doesn’t.

I use "confabulation" – despite its hint of a mental state – because it accurately describes what "AI" does – including what it does when it supplies the right answer by accident.

"AI sometimes generates fabricated responses to queries. AI scientists call these falsehoods "hallucinations," but mental health advocates suggest avoiding the analogy to human mental illness. Use the term "hallucinations" when quoting an AI scientist, but otherwise use "fabricated responses."

I'd say it's a reflection of the state of our society, when that same society will swallow hook line and sinker something that is flagrantly false, so long as it conforms with their worldview.

Slop is slop is slop is slop is slop.

"inaccurate information"

Artificial Information

May I humbly suggest the term “vomit” instead? As in, “When Mary asked Google’s Gemini the name of the state where Mar-a-Lago is located, Gemini vomited ‘Bananastan.’”

I am a real noodge when it comes to word choices. One way I have been able to rein myself in a little is a challenge I came up with: I don't get to complain about someone else's word choice unless I can suggest a replacement. Interested to hear ideas from @susanseager.bsky.social.

Dangerous

I like fabrication - confabulation is a psychological term just as much as hallucination

Perhaps 'produces nonsense'? The important part is to avoid suggesting LLMs* have consciousness, understanding or agency. *Ideally style guides should also require specificity regarding what technologies are being discussed and proscribe uncritical reproduction of the vague advertising term 'AI'.

"Working as designed but not as marketed."

If a calculator told you 2+2 = 5, you'd call it an error. If a GPS told you to turn onto a road that doesn't exist, it would be an error. When AI makes up inaccurate or just plain wrong information, guess what? That's an error. Calling it a "hallucination" humanizes a technological mistake.

Depends entirely on whether the model simply guessed wrong, or its train of thought can be shown to involve intentional misdirection. Nuance is required

Computer-Generated Bullshit or CGB.

Fabricates? Makes up?

Lies. The robot lied to you. That’s what you say.

What would you call it if the same words were found in a printed book? If it came from a database we'd call it "corrupt." The only difference with an AI is that the AI lacks reproducibility.

Output and erroneously output? These other words give it agency like it thinks when it, or the language ones at least, is basically souped-up meaningless autocompletion. It’s not lying, hallucinating, writing, stating or actively doing anything. Anthropomorphizing AI is disastrous.

Lies? Fucks up? Shits? Performs as expected?

how are y'all defining AI?

Miscalculation? Model drift or prediction error aren't single words but are accepted alternatives. Confabulation is maybe a bit better, but these are alternatives.

I know people who hallucinate and they are awesome. AI sucks.

'Hallucinate' and 'Fabricate' are equally bad. The former suggests consciousness, the latter deliberate agency. Both are wrong. 'Bullshits' and 'makes stuff up' also imply agency. Properly depersonalising the language requires two-word constructs: 'produces nonsense', 'outputs bullshit'.

I tell them they are making things up that I never used the inflection they put on me.

Wouldn't "lie" be an accurate term?