Amber Hu (they/them)No one should say that a chatbot "hallucinates". Chatbots do not have minds, they manipulate text. Hallucination requires not only consciousness but the physical brain to falsely perceive a sensation as real. Machine learning models have neither consciousness or physical form, and they never will.
Matei Matthias Covrigus š„Ø@annedrewhu
What is āconsciousnessā?
What is āconsciousnessā?
DaniĆ«l Franke š³ļøāš@annedrewhu Agreed, the "hallucination" term is just another way to make the chatbot seem more "alive" than it actually is. It's yet another way to hype it up, even when it's broken.
Panoptykon@annedrewhu great resource https://www.aimyths.org/
Siim@annedrewhu what term would you prefer? The other word that comes to mind is bullshitting but I suppose you'd suggest that also requires consciousness?
T.J. Crowder@annedrewhu Indeed. I found Ethan Mollick's article (below) useful and interesting, but took exception to his use of "lie" and "hallucinate." Anthropomorphism isn't helpful here.
(I won't say "never" to machine consciousness, but I'm certain we're a long, long, long way away from it and suspect the underlying tech will be fundamentally different.)
https://oneusefulthing.substack.com/p/how-to-get-an-ai-to-lie-to-you-in
Louise Dennis@annedrewhu Itās definitely an odd choice of language that implies some kind of victim status for the model and is considerably less clear than saying āoutputs false information ā
RRB@louiseadennis @annedrewhu I really think we should use Frankfurter's terminology: Bullshit
@rrb @annedrewhu I'd say it depends upon context. Use of of bullshit in some contexts will encourage people to take you less seriously than the person calling it a hallucination. One of the clever things about using "hallucinate" about this phenomenon is it sounds both technical and mysterious and encourages the listener to view the speaker as cleverer/more knowledgeable than them.
Ben Aveling@annedrewhu @louiseadennis @rrb I hadnāt heard the concept that LLM AI is like a machine hallucinating before, and I quite like it. (This is of course the problem with trying to suppress an idea by discussing it)
Why I like it: it makes clear that the machine isnāt lying, it just has no idea. Itās constructing a description of reality out of scraps of description, and each part fits into the next but thereās no consistency or holistic logic.
Why I like it: it makes clear that the machine isnāt lying, it just has no idea. Itās constructing a description of reality out of scraps of description, and each part fits into the next but thereās no consistency or holistic logic.
@rrb @annedrewhu In other contexts bullshit is clearly a good term for communicating the phenomenon since it succinctly communicates something about the process by which the output is generated beyond simply noting that the output is incorrect.
@louiseadennis @annedrewhu I like the term for two reasons. 1 it is accurate using Frankfurter's definition and 2 it gives the tech the gravitas it deserves.
A tool that spouts n'importe quoi just because it sounds credible is of minimal utility really. A bullshit generator
Kyle Brown@annedrewhu should the term confabulation be used instead? That describes the process better at least.
Oh.my@annedrewhu Yes! It seems to me that everyone is submitting to using the term even if they are not convinced!
Omayma (@[email protected])
I have a feeling that the term "Hallucination" of Large Language Models stuck and many people will regret it for years to come like the term AI.
Gaƫtan Perrault@annedrewhu agreed with this 100%.
Saying the Model "hallucinated" assumes that it has some normal state that isn't hallucination... Which is of course wrong.
Using this term is basically us lying to ourselves.
Damon Outlaw ļ£æ@annedrewhu the amount of arguments on the bird app of people arguing whether AI (really dislike this broad term) is sentient. Thereās arguments for, against and even loud voices bringing philosophical arguments that we canāt prove consciousness
Henry Edward Hardy
Wub-Fur Internet Radio@annedrewhu Although, like āhallucinating,ā the word implies cognition that isnāt actually occurring, I kinda think we should call it ālying.ā
Stefan Pfister@annedrewhu Are you sure?
LivePaola@annedrewhu Somebody finally agrees with you (and me): "Granting a chatbot the ability to hallucinate ā even if itās just in our own minds ā is problematic. Itās nonsense. People hallucinate. Maybe some animals do. Computers do not. They use math to make things up. [...] the term āhallucinateā obscures whatās really going on. It also serves to absolve the systemsā creators from taking responsibility for their products."
https://www.bloomberg.com/news/newsletters/2023-04-03/chatgpt-bing-and-bard-don-t-hallucinate-they-fabricate
https://www.bloomberg.com/news/newsletters/2023-04-03/chatgpt-bing-and-bard-don-t-hallucinate-they-fabricate
