ScoobyDoo27,

Well that AI is gonna be dumb as fuck

Drewsteau,

And horribly racist

sik0fewl,

Can't wait to see if it lasts longer than 16 hours - https://en.m.wikipedia.org/wiki/Tay_(chatbot)

mojo,

Oh god, people still think Tay is remotely comparable to modern LLMs.

Xeelee,
Xeelee avatar

May as well call it HitlerGPT.

fiat_lux,

It's a really dirty dataset too, with all the abbreviations and things people do to fit the character limit.

The only good use for it would be to grift the most gullible, which has historically shown to be very profitable.

Ragnell,
Ragnell avatar

Okay, even pro-AI guys have to think this is a terrible idea.

mojo,

I don’t see the issue considering other AI models have already trained off of all Twitter tweets and then a lot more. Not to defend this racist fascist, but at least he’s stating he’s doing as such instead of steering the conversation away from data scraping like OpenAI.

Ragnell,
Ragnell avatar

@mojo Fair. But man, there's a lot more BS on Twitter than there was before he took over. He's leaving some awful stuff up that even the old moderators would've deleted.

I think Musk may be doing other AI companies a favor by making it harder to data scrape from there.

TheKarion,

Audit the source code

EnderWi99in,

An LLM is only as good as the content it consumes. That's all I'm going to say about that.

stratoscaster,

“we are definitely the competition” said no competition ever lol

FelixMortane,

Not Tweets anymore.

He is training it with X-crements.

Pandantic,
Pandantic avatar

X-crement in, x-crement out.

Unaware7013,

Tay 2.0 in the making.

HiddenLayer5,

But this time the Nazi tendencies will be intentional.

ArugulaZ,
ArugulaZ avatar

That's not gonna happen, because I X'ed out my Twitter account before it became X.

promodel,

oh no!

WolfhunterGer,

So his company will create an alt-right, racist AI. Well folks get ready for some (a huge amount) of auto generated hate speech.

AWildMimicAppears,
AWildMimicAppears avatar

Finally, an LLM that does not deny output of hateful content. The amount of spelling and grammar fails in alt-right memes will go down to zero! does that mean that neonazis will become better at reading? that's bad, they might start reading books and shit.

MyDogLovesMe,

This is what Spez is feeding, maybe?

veloxy,

Remember Tay?

yip-bonk,
yip-bonk avatar

Sadly, yes.

SatansMaggotyCumFart,

Elon, stop trying to make ‘x’ a thing.

jonwyattphillips,

Tweets are public, AI chat bots are already trained on people’s tweets.

Ferk,
Ferk avatar

This is likely the reason why they wanted to tighten what you can access anonymously and instead require login, so they can try and control the rate at which third parties collect data...

yip-bonk,
yip-bonk avatar

Does one own one’s tweets xes?

Pons_Aelius,

Legally?

Yes but...by posting on the site you grant the company rights to use it as they will.

So they are technically yours.

xc2215x,

Using our tweets ? Not sure about that.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • tech
  • DreamBathrooms
  • InstantRegret
  • osvaldo12
  • magazineikmin
  • tacticalgear
  • rosin
  • everett
  • Youngstown
  • khanakhh
  • slotface
  • ngwrru68w68
  • kavyap
  • cisconetworking
  • thenastyranch
  • JUstTest
  • mdbf
  • GTA5RPClips
  • cubers
  • modclub
  • Durango
  • normalnudes
  • anitta
  • ethstaker
  • tester
  • provamag3
  • megavids
  • Leos
  • lostlight
  • All magazines