Talk:Waluigi effect

Sourcing

The Cleo Nardo post is interesting and very good... but... it's just Some Person On The Internet? I get that the arxiv paper cited it which does help, but it's also written a year ago, which is quite some time in the world of LLMs. I imagine that OpenAI would disagree with some of the more grandiose claims about how attempting to align is actually counterproductive. Can something more recent but similarly in-depth back up its claims as still relevant there?

(Also, there was an alleged quote from it before, but I found the quote nowhere in the cited source. Did the refs get swapped around and the quote was from somewhere else, perhaps?) SnowFire (talk) 05:56, 16 February 2024 (UTC)

(A year and a half later) I've removed the Nardo blog post. Per this tech.co source Cleo Nardo coined the term 'Waluigi effect'. Tech.co doesn't appear to be a reliable outlet either, and that source merely cited a lesswrong forum post, which is also unreliable.

As an aside, arxiv preprints shouldn't be cited by themselves, per WP:ARXIV, but this one has been published.

Grayfell (talk) 21:25, 3 October 2025 (UTC)

Agree, this is not a term of art, nominated for deletion 0xReflektor (talk) 10:27, 4 October 2025 (UTC)

Agree with deletion - this seems to have been a term that was only popular very briefly, and hasn't caught on. Izzy Grosof (talk) 08:02, 22 October 2025 (UTC)

can you guys bring it back? it became an inside joke with my friends, and I miss it so much. ~2026-13056-68 (talk) 19:51, 27 February 2026 (UTC)