This is a survey all #Discord users need to fill out.

[noodlemaz@mstdn.games]

Assuming it's genuine because Qualtrics isn't cheap to use - friend found it posted to Bluesky and I've seen it doing the rounds, don't know where it originated. Will update if I find out.

[andycarolan@social.lol]

@noodlemaz Done!

[bhaugland@mastodon.social]

@noodlemaz if it’s like all marketing surveys discord will ignore the results and do what they want to do.

[themnwolf@furry.engineer]

@noodlemaz ai has some very legitimate uses, none of which are being proposed to be used by discord. I've taken the survey and I've passed it on to the administration of the server that I am most active in.

[rowyn@mastodon.art]

RLB:

Me: *goes through checking all the "DO NOT WANT" boxes*
Me: *pauses*
Me: "Okay wait yes it's 2026 and you have labeled a thing AI that I have no problem with, congrats."

[noodlemaz@mstdn.games]

@bhaugland either that or they listen if it's a big negative response, nothing is lost if I register my revulsion at least.

[mgorny@social.treehouse.systems]

@noodlemaz, what should win… my hate of LLMs or my hate of Discord?

[ifixcoinops@retro.social]

@noodlemaz where'd you find this link? I can't find any mention of it on discord's website

[argonel@dice.camp]

@TheMNWolf @noodlemaz I'm neutral on transcripts of voice because sometimes a transcripts is nice and it doesn't need to be very accurate to be useful, but negative on everything else especially using user content to for model training without explicit consent. In my case that explicit response is "Fuck no, you can't train off what I say or type"

[noodlemaz@mstdn.games]

@ifixcoinops https://mstdn.games/@noodlemaz/115870988804397549

[lironah@mstdn.games]

@m @noodlemaz That one question makes this the best survey on AI I've ever seen.

[x0@dragonscave.space]

@greem @noodlemaz As a disabled person, I could see one single use for it, but AI in general, not the LLM specifically. Live captions. Voice calls with the speech captioned so that people who can't hear could participate. Speech to text is a problem much older than the LLM and uses different model architectures, and shouldn't use an LLM to do it due to halusination.

[m@martinh.net]

@Lironah @noodlemaz Also almost any poll can be improved by adding ¯_ (ツ)_/¯ idk whatevs as an option