It's not the preferred political narrative of the model that I worry about. It's how brazen they are about altering their models to achieve it. It makes me wonder what else they're altering. I have trust issues with OpenAI and Anthropic as well, but with those companies, at least I know their motives are purely profit driven. I don't have that assurance with xAI.
I don't think the comment you are replying to is giving up leverage. It's simply pointing to the OP that what he seems to be upset is not the thing itself, which everyone does, but only that he knows about it or ignores others doing the same.
When I say cancer sucks I'm not ignoring that other things also suck. So when people start trying to put those words into the words of others, they effectively act like cancer fanboys, which is weird and all about them, with no part of originating or relating to the the people they question about stuff they supposedly "ignore".
All LLMs inherit bias from their training data, and xAI’s argument is that Grok is being steered to counter that bias rather than simply inherit it. You can disagree with whether they succeed, but the act of steering isn’t automatically suspicious when knowing that every major model is steered. The relevant question is whether the steering moves the model closer to truth and neutrality, or just replaces one bias with another. At least with Grok, some of that intent is unusually explicit. I’d rather have multiple competing approaches to steering than a monoculture where every model quietly optimizes for the same idea of acceptable answers.
I don’t think that is necessarily a bad preference if this was an actual dichotomy. Not all types of manipulation is equal, and when you at least try to hide it shows at least some respect for the user.
That said, I don‘t believe this dichotomy is real. Personally I don‘t use AI, political manipulation is however only a relatively tiny part of my reasoning for opting out.
> It's how brazen they are about altering their models to achieve it.
We know all the models insert shadow prompts to nudge the answers in preferred political directions. How much more "brazen" can you get than that? Nobody is giving you fat-free results that just apply the models to your prompts.
More like a dictator versus a Poltiburo. If the “entire population” was voting on the shadow prompts of the other models they would look very different. Considering the recent election results, they would look more like Grok.
I mean one of the guys got fired by his own board for lying and is still calling the shots. Another guy sued the Pentagon during a war and we're still letting him act like a nation state.
Musk's empire of personality cult is like, idk, on slightly more cocaine?
I'm having a hard time being like: "oh, that's the bad self-appointed, self-dealing would be God Emperor. they're not all like that. why some of my very best friends are cluster B psycho con men with crime funding."
How is changing history and making a a group of objectively white people black "removing bias" ˋ? What it is is literally bias. Like thinking Pi could be 4. Removing bias ends with truth, not these crazy wonky results.
If you're aruging about historical accuracy, but still want accurate looking generated images, I don't know what to say.
But to the technical point, A large part of the training corpus has biases that if left unchecked would cause PR based disasters for the company hosting it. ie the classic black teenager/white teenager.
Now as training of models is not an exact science, and neither is the fine tuning, its analogous to forcing a water balloon into a square box. Its possible but it has odd side effects when you get to the corners.
When making a _product_ you need to choose the least worse failure case. For grok it was for a long time, pandering to the ego of the owner. For Google, who is an advertising company, its about trying not to scare advertisers. This means everthing must be vanilla
So you have a huge number of photos of white people in the training data set, but other ethnicities exist. So to make the otherwise white-biased dataset less biased, you try to e.g. add a hidden system prompt that whenever the user asks for a group of people (unspecified ethnicity), it may instead ask for "mixed ethnicities" or whatever.
Ask for a group of Nazis, and that's it - this is how models work. No "LGBTQ liberal" propaganda is needed to explain it. Unlike what Musk is doing.
That's the claim, and it's a belief that's self-fulfilling prophecy, like saying all politicians are corrupt.
If you can convince everyone that everyone is corrupt, it hurts anyone who isn't corrupt. You hear people preferring those who have no shame about their corruption, based on the premise that those who aren't overtly corrupt must be more sinister and dangerous if they hide their corruption so well.
Such a good point. Disinformation and trying to destroy sources of truth is just part of the puzzle. Often the worst damage is just "well, both sides are bad" -- because it just tells people not to listen to either side, or it is too much work.
The end goal: they want to push that you must sacrifice your rights to a monarch or authoritarian person for order and safety.
It's interesting that all models seems to be unbiased out of the box so far- that is, mostly reflecting the training data (the internet).
The whole mecha-hitler thing doesn't seem to reflect fine-tuning, it was just a prompt change.
There's been some studies that suggest that certain usage of LLMs reduces political bias, which seems reasonable. Like, how credible is climate change, are Haitians eating pets, etc. THings that have a basis in fact.
I don't put it past Elon to train a model with political bias, just that it hasn't happened yet.
> It's interesting that all models seems to be unbiased out of the box so far
This is really begging the question. If something relies on the perception of a human, it has bias. The data (or lack thereof) used to train models is per se a bias.
The mistake is assuming bias-removal is some virtuous goal to be achieved. It can't, and shouldn't. Alignment, while equally impossible, is at least a goal worth aiming towards.
I think the issue here is that there's no objective difference between "alignment" and "actively trying to shape its replies to fit a political narrative", beside the fact that the latter refers to the kind of political narratives that you don't like.
Sure, but there’s the direction of a vector, and the length of the vector.
xAI’s direction is hellish, and length is 100x any other provider’s. So, yeah, nobody is pure. But most are at least trying to be balanced and not just, you know.
That nazi salute how do you actually know it was ment to show off as a Nazi supporter? I mean who is telling this? Or is that your assumption based on something?
There is also his namesake, his racist father and his grandfathers past. There is a lot of background information available if you really look for it. And finally, if you look at story of "The Elon" and what Elon has been building, you could almost think that he is trying to actually live up to his namesake!
Trump's ex wife mentioned the only book he ever had in his bedside table at night was on hitler speeches. Multiple Trump aides have been caught reading the mEIN kampf. Stephen Miller is somehow the world's only nazi jew and writes trump's speeches.
To this cohort of people Elon spent a fortune on funding their campaign, even willing to commit election fraud (the 1 million giveaway case which is on going but seems open and close).
In front of that audience he did 2 nazi salutes chest to straight arm. He didnt apologise or explain it either.
What other possible explanation is there beyond "the dude saw a nazi adjacent politcal platform and spent hundreds of millions to make it succeed and then went mask off the second he knew there would be no repercussions"
Nah. My beliefs are actual truth, so if provider is shaping their models according to my preferred political narratives that's correct and only moral thing to do. Anything else would be morally bankrupt.
Who or what is this latter "globalist" view? Like it's such an oversimplification of multiple, independent actors' acts that it's just a plain bad world view/model. What would be the common goal? You need some extra amount of tinfoil for such bullshit.
No, but technically, all models go through an alignment phase where you feed data that's aligned to your goals and values to train the model so that it will exhibit the kind of behavior you want.
There's no "politically neutral" value system anyway.
Doesn't mean we shouldn't try to make the models more inclusive, less biased and less prone to extremism, but in a technical sense yes, actually everyone does it.
And it's all the same to you? You don't care which values those things have? Of course there will always be underlying values. I wouldn't go as far as calling everything political narratives.
I think it's just pretty clear that Elon's values are not what most people want the world to be shaped by.
That's reductive, how many other models had a mecha hitler incident? Or a "let's talk about white genocide in south africa" incident?
There is some truth to what you say, but most model providers I would say are engaged in CYA type shaping moreso than anything, grok is actively and openly being developed to spread a white nationalist agenda. There are levels to this.
False, most are just throwing in data into the pipeline and hoping something very smart comes out the other end. Their shaping is to have it code very well, beat all the benchmarks and be safe enough that the powers that be don't and public sentiment doesn't turn against them.
Sure, every author has a bias. But a fair selection of human written sources will be pretty balanced (of course, given the ratio of languages, surely western models will have a western-christian bias - presumably Chinese models less so, but this latter I have no way of checking).
If you train specific stuff on top, or deliberately filter the sources (e.g. Tiananmen square), your model is deliberately less honest on that topic. Grok is probably the worst in deliberately filtering and training "out" specific stuff (to the point where Elon posted stuff that "they will 'fix' the models" real output when it said something true but remotely liberal). Claude and chatgpt definitely have some similar stuff, but mostly to protect themselves (e.g. suicide prevention, not saying slurs, etc). I don't think the two is comparable (reality bending vs basic etiquette-kind of not saying everything out loud)
Yes. But only one trillionaire is screaming from twitter that he will use his wealth to combat the woke mind virus by tuning his model. So, ya know, there’s THAT.
Yeah but their political narratives involving dehumanizing people, and hurting minorities - which is way worse. The attempt at false equivalence between traditional American propaganda and right wing American propaganda is disgusting.
Maybe, but some political narratives are good and right and true, and others are bad and wrong and false. (I am not joking: objective reality exists and most of politics isn't subjective.)
It has occurred to the OP, but those ones are ok as they fit OP's political leanings. They just want to silence other political narratives that dont agree with theirs, they want Grok in the Gulag.
No. Other companies are altering thier products to chase market share and profits. Grok is run by someone beyond profit motives who is actively promoting a personal agenda. I would rather work with Microsoft's counsel of dark wizards than an individual with an axe to grind with the world.
From a business perspective, a company that trains it's LLMs to having boring, mainstream, generally-inoffensive views is a big selling point over whatever the hell Elon is doing.
Drugs. Drugs are what Elon is doing. And it's pretty cringe.
But it's remarkably similar in cringe to that little "secret erection" look Amodei gets when he talks about millions of unemployed people, or or Altman rolling through Pacific Heights in a four million dollar Swedish hypercar holding the steering wheel wrong the day after yet another lecture about UBI.
You only get to become one if the idea of spending 60 hours a week typing code into a computer during the late 90's/early 2000's is palatable to you. This causes a dweeb selection funnel.