The Nature of Hallucinations • Hacker News Board

Re: The Nature of Hallucinations

#11

Post by s-macke » Tue, Sep 23, 2025, 2:08 PM UTC

Panzerschrek wrote: post #5

I find the term "hallucination" very misleading. What LLMs produce means really "lie" or "misinformation". The term "hallucination" is so common nowadays only because corporations developing LLMs prefer using it rather than saying the truth, that their models are just huge machines for making things up. I am still wondering, why there are no legal consequences for authors of these LLMs because of that.

Author here. The discussion about this wording is actually the opening section of the article.

> Unfortunately, the term hallucination quickly stuck to this phenomenon — before any psychologist could object.

Re: The Nature of Hallucinations

#12

Post by add-sub-mul-div » Tue, Sep 23, 2025, 3:04 PM UTC

partomniscient wrote: post #3

Title should be amended to "Nature of AI Hallucinations". The first line "Why do language models sometimes just make things up?" was not what I was expecting to read about.

It's probably futile by now to fight that "hallucination" and "slop" have become synonyms for AI output and the AI context will be their most common or default use going forward.

Regardless of whether those terms in the AI context correlate perfectly to their original meanings.

Re: The Nature of Hallucinations

#13

Post by partomniscient » Sun, Sep 28, 2025, 5:05 PM UTC

add-sub-mul-div wrote: post #12

partomniscient wrote: post #3

Title should be amended to "Nature of AI Hallucinations". The first line "Why do language models sometimes just make things up?" was not what I was expecting to read about.

It's probably futile by now to fight that "hallucination" and "slop" have become synonyms for AI output and the AI context will be their most common or default use going forward. Regardless of whether those terms in the AI context correlate perfectly to their original meanings.

It's futile in the sense that absolutely anything and slop are synonymous with AI output.

There was a post somewhere about the irony of a human having to moderate resonable AI slop vs. shithouse AI slop.