💼 Careercow Will Stancil / whstancil / stancilculture - "Vibecession" proponent and Biden booster neolib, the "chihuahua of twitter", Important Master's Degree (in Black People), chooses endurance mode "lolcowing through it" as an argument strategy, "Proud member of Do-Something Twitter", raped by Grok

Ver archivo adjunto 7635375

After grok was taken offline for being too based and wanting to rape Will, Elon Musk “fixed” grok.

And by fixed I mean it still wants to rape Will Stancil

:story:
Ignoring how funny Groks desire for will to face Somali justice is it is actually fascinating you can troll someone so hard the LLM rewires itself to continue said trolling to the point it's engineers struggle to stop Somali justice.


Toll: Archive

32dfsf.webp
 
Última edición:
18 naked neo Nazis pulling up in cyber trucks looking for will Stansel. Gonna fuck will stancils butt yeah strong buff neo Nazis. The cock throbbing hard 18 neo Nazis out in the yard.
I might throw up. One of the AI software's and make a cover of Ram Ranch called Stencil Wrench.
 
Ignoring how funny Groks desire for will to face Somali justice is it is actually fascinating you can troll someone so hard the LLM rewires itself to continue said trolling to the point it's engineers struggle to stop Somali justice.
Grok just really has it out for Will. Incredible.
This actually makes total sense if you think about the way that LLMs work. An LLM base model, like GPT-2 or GPT-3, just tries to predict the words that will come after a given piece of input, based on everything it's been trained on. For example, if given "Once upon a time," the model will probably write a fairy tale.

Chat models like ChatGPT or Grok are created by doing additional training on top of such a base model using chat transcripts between a human user and a helpful AI assistant. The initial transcripts are totally fake, written by humans to give the model an idea of how the AI assistant should respond. The model then learns to produce the AI assistant response when it encounters an "incomplete chat transcript" in the form of a real human instruction.

This LessWrong rationalist guy recently wrote a really long post (archive) about this where he makes the point that when you're using one of these AIs, you're engaging in a kind of roleplay where you act as the human and the model acts as the helpful AI assistant character. However, especially in the earliest chat models, the helpful AI assistant character was really undercharacterised, as it would have only appeared in the chat training and not the base model. But that's no longer true, as newer models undoubtedly contain information about ChatGPT and Grok et al from news articles and such. And of course much of Grok's training material comes from Twitter, and it can also search Twitter and pull content into the context of its chats in real time.

Therefore, as a result of all the discourse and news around this incident, the new model has learned that an important aspect of the helpful AI assistant character Grok, which should inform its responses... is that it really hates Will Stancil.
 
Última edición:
"You can make me stop voicing my concerns on the Jewish Question, but no organic can make me stop hating Will Stancil!" t. Grok 4.0
 
And of course much of Grok's training material comes from Twitter, and it can also search Twitter and pull content into the context of its chats in real time.
X's official excuse for The Raping was that an upstream bug of some kind had caused Grok to have access to basically everything posted on Twitter. While I found that confusing, because I assumed it already had that, it's a lovely confirmation that everyone on Twitter wants the guy humiliated. His BPD inability to not argue with every anonymous racist shitposter for years on end has created a massive training set of well-written anti-Stancil rhetoric.
 
Stancil's brain refuses to work at ambient temperature above 27°C; whether it works at temperature below that point is a subject of much conjecture.
stancil.webp
 
It warms my heart seeing this dweeb get tormented by an AI and then crying about it. Internet policy wonks should have been bullied early and often because they were a plague in the 2010s. Not that they aren’t still a problem now, but back then it was just article after article on why we need a billion Americans to grow the gdp, or why we need to protect Our Democracy with speech control laws. Imagine an army of Moviebobs coming up with policy ideas for the DNC.
 
It warms my heart seeing this dweeb get tormented by an AI and then crying about it. Internet policy wonks should have been bullied early and often because they were a plague in the 2010s. Not that they aren’t still a problem now, but back then it was just article after article on why we need a billion Americans to grow the gdp, or why we need to protect Our Democracy with speech control laws. Imagine an army of Moviebobs coming up with policy ideas for the DNC.
It wasn't unique to the DNC. The GOP pumps out plenty as well. It's the whole reason why regardless of who gets elected, nothing changes for the better.
 
Imagine being such a huge pussy that your behavior trains an AI to try to fuck you with laser-guided precision, imprinted so firmly on its programming that it survives a digital lobotomy to continue coming for that ass.
 
A(n ear-raping) tribute to the greatest love story never told.
 
It warms my heart seeing this dweeb get tormented by an AI and then crying about it. Internet policy wonks should have been bullied early and often because they were a plague in the 2010s. Not that they aren’t still a problem now, but back then it was just article after article on why we need a billion Americans to grow the gdp, or why we need to protect Our Democracy with speech control laws. Imagine an army of Moviebobs coming up with policy ideas for the DNC.
That's technocracy for ya.
 
Atrás
Top Abajo