[ExI] i don't know about llms, but I am feeling the pain was: RE: Do LLMs feel pain?

[spike at rainier66.com]

…> On Behalf Of Jason Resch via extropy-chat
Subject: [ExI] Do LLMs feel pain?

 

https://arxiv.org/pdf/2609.16247 

 

This paper identifies a reliable pain axis across various LLMs, which activates under certain conditions:

 

See page 13 for examples. Treat your AIs right!

 

Jason

 

 

 

 

 

Jason, this article is disconcerting for reasons unrelated to its subject.  It was a cool idea and all, however… there is a dead giveaway that this paper was mostly AI generated, with little human input.

 

Hint 1: it is on page 5.

 

Several of my friends over the years have flattered the hell outta me by asking me to proofread their PhD thesis.  So I do.  I learn a lotta good stuff, such as how to spot bugs in research papers.  Amara Graps’ paper, oh that was cool.  She’s alive and well btw.

 

This paper I would send back for cleanup.  A lotta cleanup.  

 

Hint 2: right at the bottom of page 5.

 

Reasoning: every research paper is filled with yakkity yak and bla bla, but when the equals sign drops, the bullshit stops.

 

If a researcher puts any equals sign in her research, she is making a strong claim: this stuff IS that stuff.

 

Strong claims require strong evidence.  OK so that equation.  Prove it.  Even if you read down to there, do you know what that equation is claiming?  Do you know what all those variables are for?  Why don’t you know?  Because the authors didn’t define them.  Any variable used in any equation in any credible paper must be carefully unambiguously defined before it can go in any equation.  It is analogous to a futures bet: everything must be unambiguously defined, a metric of some kind, no squishy adjectives allowed.  You may not even reference down to an appendix.  It hasta already be defined, ideally right there with the equation.  I see no definition of the variables anywhere in that paper.  Bzzzt.  Mostly AI generated.  Out it goes.

 

I started this commentary with the comment that the paper is disconcerting.  Reason: it is getting damn hard to distinguish between fake research, presumptuous research, mostly AI generated research, good research (four different things.)  I was reading this paper all the way down to the bottom of page 5 before I was sure.  It looked like real research up until that equation.  

 

The body of recent scientific literature is now probably filled with AI generated stuff, which is going into AI training research.  Not only are we breathing our own fumes, AI is breathing its own fumes, which previously whoofed ours.

 

spike

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20260924/a0e0804b/attachment-0001.htm>