Oh yeah, you're basically having a conversation with the Hegemonic Culture.
You can try and undermine it by priming it with leftist connections; Asking if it's familiar with leftist theories, authors, etc.
this post was submitted on 20 May 2025
76 points (97.5% liked)
chat
8407 readers
154 users here now
Chat is a text only community for casual conversation, please keep shitposting to the absolute minimum. This is intended to be a separate space from c/chapotraphouse or the daily megathread. Chat does this by being a long-form community where topics will remain from day to day unlike the megathread, and it is distinct from c/chapotraphouse in that we ask you to engage in this community in a genuine way. Please keep shitposting, bits, and irony to a minimum.
As with all communities posts need to abide by the code of conduct, additionally moderators will remove any posts or comments deemed to be inappropriate.
Thank you and happy chatting!
founded 3 years ago
MODERATORS
Exactly, a useful technique (and maybe where The Rehearsal S2 is headed) is to tell the LLM that it's playing a defined role.
"You are a worldbuilding assistant that provides suggestions for and keeps records of the world I am building. You are a scientific socialism expert and you guide worldbuilding using analysis of economic system dynamics instead of idealistic views. "
Yeah, my general prompt template is something like...
- You are an expert at whatever, with a passionate background in this and that.
- Then state your information and what you wanna do with that information.
- then something along the lines of: take a moment to step back and visualize the challenge I've presented you with. Run through it over and over in your mind from different angles until you've found a solution you're happy with. Use best practices or whatever.
- if you have any questions or need something clarified, please ask.
Gotta remember that regardless of whether it's intelligent or not, it's largely attempting to emulate human text response. So if you want to elicit a good response from it, use the techniques that get the best responses from real people.
Yep. You have to give them a persona
It's just Omelas. Can't just have a good thing, there has to be a bad thing.
Omelas is both a criticism of Utopian beliefs, a criticism of those criticism, and a liberal capitalism indictment.
The idea that deep seek would somehow have an actual leftist bent has been the most ridiculous thing many of you people for some stupid reason actually believe.
Some people seem to accept the imperial media’s premise, consciously or not, that Deepseek is just a black box with a CPC agent inside answering questions and monitoring prompts.
Anytime you leave things to the "imagination" of the AI it will always pick the most generic, cliche and unoriginal things possible. That's why it does things like this. The only good thing AI can do for worldbuilding is getting a response from it, and then doing the opposite.
The new boss can never be different from the old boss. 
It would seem Deepseek did not get the memo about the adult brainwashing app I'm making on behalf of President Xi. 
Historically accurate for almost all societies though.
Chatbots are only as good as what they're trained on, its mostly Western brainworms so don't expect differently. Only positive of Deepseek vs western chatbots is it has a better grasp of what diamat and dialectics are outside that common but purposefully limiting miseducated hamstring of synthesis, thesis, antithesis.
I figured this out when I asked Deepseek about fascism and so-called neofeudalism, while being generally a radlib it went full 'see Uighurs China bad' in said chat, which was a big nope moment. Also it presented Orwell as a positive figure, all I need to know about trusting this lil bot. Good for superficial surface level shit, if you're looking for comrade chatbot you won't ever find it unless you look in the mirror.
Tweak your prompt a little. Ask it to not ascribe any nefarious or negative purposes. Draw from Incan Empire history (they did similar stuff).
But yeah that sort of stuff and the “kill the boer incident” just makes it very obvious you can’t bother it certain things.
A while back I played around with chatGPT trying to write random dystopic stories. Every other plot it inserted "a group of plucky individuals and intelligence officers" to save the day. The CIA is your friend.
But seriously, deepseek and chatGPT are at best a lossy compression of the text distribution used to train it. It will necessarily mode collapse on westoid Anglo thrash when writing in English. As another user put it, "why would I want to read your LLM 10000 word story if I can just ask another LLM to summarize it back to the 30 word prompt?"