News for hackers. Stuff that matters.

A comment on Overcoming the Limits of Large Language Models. Parent comment

Re:Overcoming the Limits of Large Language Models by jwuphysics on Wednesday July 17, 2024 @11:00PM (#40992115)
> Why not just say "incoherent output"? Because the biggest problem with hallucinations is that the output is usually coherent but factually incorrect. I agree that "hallucination" isn't the best word for it... perhaps something like "confabulation" is better.
[ Parent | on HN ]
Re:Overcoming the Limits of Large Language Models by nl on Thursday July 18, 2024 @12:14AM (#40992460)
And we use "hallucination" because in the ancient times when generative AI meant image generation models would "hallucinate" extra fingers etc.

The behavior of text models is similar enough that the wording stuck, and it's not all that bad.

[ Parent | on HN ]
Re:Overcoming the Limits of Large Language Models by darby_nine on Friday July 19, 2024 @08:17AM (#41005615)
That was a misnomer—hallucination refers to perception, not generation. Completely misled an entire generation of people.
[ Parent | on HN ]
Re:Overcoming the Limits of Large Language Models by Raphaellll on Friday July 19, 2024 @11:49AM (#41007681)
"hallucination" was coined in the context of text generating RNNs. Specifically in this blog post by Karpathy in 2015: https://karpathy.github.io/2015/05/21/rnn-effectiveness/
[ Parent | on HN ]
Re:Overcoming the Limits of Large Language Models by linguistbreaker on Wednesday July 17, 2024 @11:44PM (#40992307)
I appreciated a post on here recently that likened AI hallucination to 'bullshitting'. It's coherent, even plausible output without any regard for the truth.
[ Parent | on HN ]
Re:Overcoming the Limits of Large Language Models by wkat4242 on Thursday July 18, 2024 @02:31AM (#40993013)
While I have absolutely no issues with the word "shit" in popular terms, I'd normally like to reserve it for situations where there's actually intended malice like in "enshittification".

Rather than just an imperfect technology as we have here.

Many people object to the term enshittification for foul-mouthing reasons but I think it covers it very well because the principle it covers is itself so very nasty. But that's not at all the case here.

[ Parent | on HN ]
Re:Overcoming the Limits of Large Language Models by pessimizer on Thursday July 18, 2024 @01:52PM (#40998106)
"Bullshitting" isn't a new piece of jargon, it's a common English word of many decades vintage, and is being used in its dictionary sense here.
[ Parent | on HN ]
Re:Overcoming the Limits of Large Language Models by MattPalmer1086 on Thursday July 18, 2024 @07:09AM (#40994472)
More true to say that all output is bullshitting, not just the ones we call hallucinations. Some of it is true, some isn't. The model doesn't know or care.
[ Parent | on HN ]