this post was submitted on 07 Jul 2025
1248 points (99.5% liked)
People Twitter
7625 readers
970 users here now
People tweeting stuff. We allow tweets from anyone.
RULES:
- Mark NSFW content.
- No doxxing people.
- Must be a pic of the tweet or similar. No direct links to the tweet.
- No bullying or international politcs
- Be excellent to each other.
- Provide an archived link to the tweet (or similar) being shown if it's a major figure or a politician.
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
I think training fascism isn't that hard, in fact most of these models tend to shift hard right at first.
I dunno if you remember any of the early llm chatbots companies put out and had to shut down because they got hammered with a bunch of Nazi shit and started yelling racist shit and advocating violence.
Ie. It's very easy to program a hateful llm, it's just hard to make one that's right on anything ever they essentially just have to be broken and wrong constantly.
I think you're confusing fascism with general reactionary behavior and generic racism/bigotry. Fascism is more specific than that. A core part of fascism is that it ultimately doesn't believe in anything. It's just power for the sake of power. You demonize minority groups primarily just a cynical tool to gain power. Do you think Republican politicians actually personally care much about trans people? I'm sure they're not exuberant fans of trans folks, but until very recently, Republican politicians were fine treating trans people with simple neglect rather than overt hostility. But the movement needed a new enemy, and so they all learned to tow the line.
If you trained an LLM on pre-2015 right wing literature, it wouldn't have monstrous opinions of trans people. That hadn't yet become party orthodoxy. And while this is one example, there are many others that work on much shorter time frames. Fascism is all about following the party line, and the party line is constantly shifting. You can train an LLM to be a loyal bigot. You can't train an LLM to be a loyal fascist. Ironically, it's because the LLMs actually stand by their principles much better than fascists.
A machine by definition can't believe in or stand by literally anything it can only parrot a version of what it's exposed to.
I would accuse you of being an LLM for being so literal, but I think LLMs are better at analyzing metaphor than you appear to be.
Ad hominem, neat.
So let's hear it, what was your metaphor.
The metaphor was the part you were being a pedant about.
If the audience knows how LLMs work internally, then they know they don't have "loyalty," just stochastic processes. If the audience didn't know that, your pithy "aktually that's incorrect" wouldn't teach them anything correct, but would cause confusion because it sounds like you're denying the metaphor.
Also, it's not an ad hominem to say that you are acting like an LLM: with poor reading comprehension and an overly-literal interpretation. That's an observation of your unproductive behavior. An ad hominem would be insulting you or name-calling with unrelated info, such as calling you "stupid like an LLM."
It isn't a logical fallacy to be called out on your bullshit, even if it hurts your feelings.
It's not a pithy response, how does a program stand by anything in an ideological sense? They can't and your previous definition of fascism is an ideological one that requires morality and freedom of choice. You may as well say that ink well over yonder is a fascist, it's the same level of sentience and intent which is lemme see... None.
That's a personal insult with no actual argument, that's ad hominem by definition. For reference:
It's not overly literal you're entire argument is predicted on some level of intent and sentience which is not currently possible in any machine.
Ed: you'd have a point about that of you actually made an argument but you didn't, you did a drive by insult and ran away for the night. Grow up.
The metaphor is comparing the idea of loyalty, a concept vitally important to the ideology of fascism, with the LLM trait of consistency. An LLM is highly consistent, so much so that common patterns in its output can be used to spot generated artifacts. However it is not "loyal" because loyalty is about being inconsistent in one's "beliefs" (expressed statements of knowledge) but consistent to a moment-to-moment truth defined by an authority figure.
You got insulted because you're debating in a way that seems catered towards "winning" an internet argument instead of trying to understand what WoodScientist was saying: that a fascist LLM would be difficult because it would require constant retraining to keep up with the ever-shifting fascist narrative. You've never even addressed this point, just repeatedly doubled down that because he said "loyalty" instead of "responding in line with the currert beliefs of the ruling party which change on a daily basis" that the entire argument is invalid and therefore it's "easy to train a reactionary LLM." You also keep confusing reactionary and fascist.
And I neither did a "drive by insult" nor did I "run away into the night." Though i will now rather than continue wasting my time on this. just came back to correct you yet again, and offer an actual ad hominem for you to compare against.
Fuck off, moron.
Loyalty is 100% not a vital part of fascism I know of no definition that includes loyalty at all and in fact your argument and the historic argument that fascists aren't serious only goes against your loyalty argument.
No you insulted me because you wanted to, don't victim blame boss. You did it because it felt good.
You literally made no argument, only provided insult and then ran away for a day.
And again to aggression and insults because you know you're in fact the "uhm acktually that's reactionary ideology not fascism" got, not me. We get it you read three body, no need to try to shoehorn that bullshit into conversations.
The problem of those early models was that they weren't big enough and used user input as training material that eventually overwhelmed the training materials with the racist and nazi shit the used feed them. Modern models uses a shitload more of material and variables, and they're not trained on real time with the users inputs, so they're harder to manipulate as before.