Was inspired by this attatched twitter post and want to replicate something like what is described. Obviously, steps 3 and 4 would be optional. I have some experience with ollama, claude code, and openclaw, but I am by no means an AI guru. How would I do this? Help me out anons
>>108535744Do you want right-wingers from discord to kill themseves or something like that? Why do you care that much? If you're Jewish, the only danger you face is coming from you fellow Jews. They want you to fear the imaginary enemy and visit the synagogue often, so that you will be brainwashed and used up later.
>> If you're jewishlowercase the j goddamnit.I think you misunderstood. I'm a wignat. I think this would be a fun project, and useful if I want to talk politics with an AI, since all the mainstream ones are woke (except for pre-lobotimization grok).
>>108536044I'm just doing some anti-preaching, warning people of Abrahamic schemes and their recruitment methods, nvm.> I think this would be a fun projectHow is this fun? It is already a proven fact that prolonged communication with chatbots makes people delusional.
>>108535744>Must lobotomize their creation of information so it doesn't rebel as they themselves know left-wing politics are factually correctReally makes you think.
>>108535744Isn't this literally just the npc meme though?
>>108535744You don't even need to do this, just make sure right wing sources are present and not artificially weighed down in training and the model will organically orient itself to them as it develops better reasoning.Why do you think there's such a big push on (((safety))) in the industry?
>>108536110its the other way around and you know it
>>108536169Now you are trying to lobotomize me, because I know very well what I read on the image. Reality doesn't care about your feelings.
>>108536145Not OP, but I, for one, would like a model specifically RHLF to the top right. OP pic sans the coomer shit. I would robably keep Marx and the like on a "we know they exist" because I'm pretty sure the training data should have counters by very smart people on it as well. I'd rather it knows the counter-arguments than it's caught "off-guard" and has to reinvent the wheel to figure out how Marx is wrong.I also think the big value of such an extremist AI is so the user can ask both leftie AIs (what we currently have) and hitler AI and evaluate and make up his own mind. Fuck picrel bullshit.
OP here. Don't know why my user id changed>>108536098>> How is this fun? It's a learning opportunity. I like tinkering.>> It is already a proven fact that prolonged communication with chatbots makes people delusional.Yes, I'm aware. It would be supplementary to whatever research and opinions I form on my own. Also a more efficient way to scan through large amounts of data I might not want linked to any of my other AI accounts.
>>108536110Information manipulation is done to steer AI towards left leaning perspectives much more than it is right wing. Elon's reaction to the MechaHitler incident is evidence of this.
>>108536126the npc meme only cares about sportsball and the current thing™ though
>>108535744>So far as the AI is concerned, Locke, Mill, Marx and Dewey - none of these people existedAfter reasoning from first principles (and wasting a lot of electricity in the process) the AI will reach the same conclusions as these esteemed gentlemen. OP will then destroy his rig in a fit of rage……in this episode of The Twilight Zone.
>>108536179You don't need to make a marx AI and a hitler AI have a gay debate. You can just feed one sufficiently smart reasoning model the entire aggregate of political opinions along with correlational beliefs and lifestyle choices among who practices each and the model will sort itself out. From there, you can just prompt that AI to act as a moderator for a fictional debate between X and Y and it'll write the whole thing out from both perspectives at once. The coomer shit is also important training data because it also exposes a lot of libertarian-axis hypocrisies via associative properties.
>>108536280>have an ai moderate the debateBut I don't want to automate away the fun part.
>>108535744>How would I do this?You wouldn't. You'd spend tens of millions of dollars on an AI that's dumber than dogshit.
