Molly WhiteHave you ever wanted to start editing #Wikipedia, but got overwhelmed or felt like you didn't know where to start? Every time I encourage people to start editing, I hear that, so I'm trying to help.
YouTube
this was inspired by some conversations on bluesky, where people have been concerned that #Wikipedia might become overwhelmed by AI-generated #enshittification. https://bsky.app/profile/molly.wiki/post/3km4ejpq7kt2s
if that worries you too, the number one thing you can do is learn how to edit.
Andrei Kucharavy@molly0xfff I've been having an edit there and back on the LLM page to pin the origin of the name and the definition of the size, that most researchers agreed on as of mid-2022.
It's been now edited back ~ 3 times to praise Transformer, Google, and OpenAI.
It's been now edited back ~ 3 times to praise Transformer, Google, and OpenAI.
@andrei_chiffa at that point it's best to bring it to the talk page for discussion rather than editing back and forth
@molly0xfff did that, got encouraged to post by the same editor who reverted the edits; came back to ask what happened the first time without knowing who reverted it; got suggested to add more; edits got reverted a second time. Checked a 3rd time to see if it was not an error; still got reverted.
@andrei_chiffa got a link?
Here is the talk page:
https://en.wikipedia.org/wiki/Talk:Large_language_model#Source_of_the_term_and_disambiguiation
(andrei.chiffa is me, Dancing Philosopher is the editor)
Here is the diff of the part I added:
https://en.wikipedia.org/w/index.php?title=Large_language_model&diff=1164830127&oldid=1164685868
Here is the diff where "Dancing Philosopher" removed it.
https://en.wikipedia.org/w/index.php?title=Large_language_model&diff=1167234405&oldid=1167118942
Failing to mention AI2 and ELMo both shifts the credit to Google and OpenAI and erases the reason all early LLMs had muppet names; whereas the Stochastic Parrots erasure hides the true scope of issues with LLMs beyond AGI.
@molly0xfff And whitewashes / browashes the field of LLM security and safety, because neither @timnitGebru, nor @emilymbender fit the narrative.