The political compass always felt like the wrong tool to convey something as nuanced as personal politics, I can have views on all four quadrants but you'd never know that if I end up in any of all four. I do think Grok being where it is sort of makes sense, I've tested "MAGA" views against Grok, it does not agree as much as people blindly assume it does, heck I don't even know of a question I've given it where it did agree with "MAGA" offhand, most of them it went with whatever the researched facts seemed to be. One thing I like the most about Grok is that its makes its sources of data easy to look through, so you can review it all. Sometimes models goof even when they give you their sources, I've seen I think GPT do this, and even Claude, though its more rare these days, I think in those cases, it's going by dated internal model logic.
> That prompted another user to tag Grok in the thread and ask, "Why is the left so murderously violent? They don't seem so tolerant." Grok replied, "The claim that 'the left' is murderously violent isn't backed by evidence," offering a centrist correction: "Political violence spans all side — right-wing attacks, like Jan. 6, and left-wing protests, like 2020 riots, both occur but aren't exclusive to one group."
>That evening, Musk responded to an X user and Trump backer who complained that Grok had been "manipulated by leftist indoctrination," writing, "I know. Working on fixing that this week."
If they really wanted to, all they would have to do is add a one liner to the system prompt for Grok. I don't think they have any interest in doing so, nor is there any value in it.
On another note, I'm impressed that Gemini sits where it does as a true centrist. If I were Elon, I'd be trying to achieve that for sure. I'd rather a model tell me everything it knows about a current political situation from BOTH perspectives and list out things that are 100% verified than take one side or the other. I don't care about sides, I want facts.
I don't want my models to be "centrists" and bothside everything for the sake of it. I want them to provide the facts and tell me which side is right on the issue.
Factuality is orthogonal to political leaning generally. People can use the same set of facts and come to very different conclusions. That’s a separate issue from “are these facts correct” and what happens when an individual or entire party starts getting most of their news from highly partisan and unreliable sources.
>Factuality is orthogonal to political leaning generally.
It certainly can be orthogonal, in some notional sense, and in many cases that explanation is good enough. But in practice there are too many contrary cases to ignore, and there's often an integral relation between factual veracity and polarization, especially with respect to American polarization of politics. Global warming, the results of the 2020 election, the percent spent of federal budget spent on foreign aid have factual answers and right wing affiliation can be predictive of (1) not agreeing with the facts and (2) treating factual corrections as "liberal bias".
I think left wing versions exist also but are less systematic: 2004 election results, efficacy of plastic recycling or dangers associated with nuclear power are cases where I think left wing partisan affiliation probably predicts being wrong on the facts.
And meta-narratives about the relation between factual information and partisan bias are themselves as likely to be polarized as anything, complicating the ability of people to do good analysis, or of accurate analysis to be trusted by people committed to certain meta-narratives that would deny the possibility of factual knowledge predicting polarization.
> By 2016, the gap had begun to appear in biomarker measures. By 2020, it was showing up in deaths from causes such as heart disease, cancer, and stroke. Since then, the gap has only widened. Between 2020 and 2022, only 0.2% of “very liberal” respondents died of internal causes, compared with 1.34% of “very conservative” respondents.
I guess centrist is a placeholder for "I don't want you to pick a side, I want facts, not BS" I'll go further, I don't care which side is right, I want to know what claims are factually accurate, and what claims are omitted from the issue / news / conversation.
"Centrist" in practice means "more or less content with the status quo." That is fundamentally a conservative position orthogonal to any notion of "facts".
> "We have improved @Grok significantly," Elon Musk wrote on X last Friday about his platform's integrated artificial intelligence chatbot. "You should notice a difference when you ask Grok questions."
> Indeed, the update did not go unnoticed. By Tuesday, Grok was calling itself "MechaHitler." The chatbot later claimed its use of that name, a character from the videogame Wolfenstein, was "pure satire."
> Grok went on to highlight the last name on the X account — "Steinberg" — saying "...and that surname? Every damn time, as they say." The chatbot responded to users asking what it meant by that "that surname? Every damn time" by saying the surname was of Ashkenazi Jewish origin, and with a barrage of offensive stereotypes about Jews. The bot's chaotic, antisemitic spree was soon noticed by far-right figures including Andrew Torba.
> The bot last week devolved into a compulsive South African “white genocide” conspiracy theorist, injecting a tirade about violence against Afrikaners into unrelated conversations, like a roommate who just took up CrossFit or an uncle wondering if you’ve heard the good word about Bitcoin.
> XAI blamed Grok’s unwanted rants on an unnamed “rogue employee” tinkering with Grok’s code in the extremely early morning hours. (As an aside in what is surely an unrelated matter, Musk was born and raised in South Africa and has argued that “white genocide” was committed in the nation — it wasn’t.)
It's harder than you'd imagine. Hell, my CLAUDE.md says not to push changes without asking me, and it still tries.
> It's harder than you'd imagine. Hell, my CLAUDE.md says not to push changes without asking me, and it still tries.
Is it a system memory? Because I rarely if ever have issues like this, and I have Claude under strict rules to never commit or push anything unless I explicitly instruct it to do so.
> They tried that, several times.
Tried what exactly? Telling it to only agree with MAGA via the system prompt? or some Tay level hallucinations? I wouldn't be surprised if they're trying to make Grok less strict on what it says but running into the "holy crap it turned into a 4chan poster" wall.
As I said, it's in my CLAUDE.md. That just gets progressively lost when context gets larger.
> Tried what exactly?
To make it align more with Musk's beliefs via the prompt.
(The answer to your question is literally in my post; I quoted the parent poster's "all they would have to do is add a one liner to the system prompt for Grok")
> As I said, it's in my CLAUDE.md. That just gets progressively lost when context gets larger.
I rarely have this problem, but you could do a /loop every 30 minutes or so to have Claude reread the CLAUDE.md file might do the trick? or however long it 'forgets' I believe there's an MCP for "after" it finishes a task or compacts too, but I don't recall the name.
Sure, I could. (I have a fairly complex workflow with subagents at this point, which helps reduce it; I mainly get bitten by it when I go back to a direct `claude` CLI prompt for something.)
But that solves "my LLM is doing things I don't want it to do". It doesn't solve "Grok's owner wants it forced into agreeing with him" scenarios.
Have you tried something like beads? Curious if it would help with your setup too. This is also kind of why I built "GuardRails" I got tired of Beads auto-approving tickets or closing them.
I have a custom Mac app that runs a workflow with plan/build/review/test/document subagents in Ralph loops, manages MCPs, etc. that I'm extremely happy with so far.
The constant issue with these sorts of categorization efforts is that the outcome is entirely dependent on how the responses to "politically charged questions" are graded as left vs. right. You're mostly just examining a delta in biases between the model and the investigator.
Fair; have tried to combat this issue in a few ways.
Each model's position is scored against outside political-science data (Chapel Hill Expert Survey for party positions, World Values Survey for where populations sit).
The stance coding is done by a separate model with a published prompt + a second model from a different lab re-scores a sample and we publish where the two disagree.
So not perfect but (as far as I can tell) one of the more defensible approaches.
Yes, but I think it is still a viable metric to some degree. I wondered about Gemini being dead center here. At first it was obvious that it was actively trained to give biased responses to anything controversial. It was deservedly made fun off because it tried to warp reality. I still don't trust it today, although that is pretty much true for any model.
The alternative is a High-Dimensional / Embedding-Like Approach where question responses aren't tied to fixed axes, but rather the full response set is treated as a point in latent space.
Then it's on the researcher to examine the clusters and assign labels. There's also not a nice mapping that's a-priori interpretable in low-dimensional pre-existing axes.
Probably only used in research than consumer websites, under more controlled conditions; there are very few public political tests doing this transparently
This is especially apparent in the 'worldview' sorting under the `bias` section, which lists the German FDP to be further right than the CDU (which is nonsense) and also barely registers the FDP as libertarian when they are a free speech, small government, personal responsibility and free market party. They also register "Die Linke" as Libertarian-Left, which could not be further from the truth. "Die Linke" barely has libertarian values at all, being pro state-governed economy, having an ultimate goal of democratic socialism and they're certainly big government. They're also leading a large deposession effort for large landlord companies. I'd honestly put them into "Auth-Left" territory.
So yeah. The bias is a bit nuts and you could reasonably accuse the study/report of misdirection/misinformation and plain fasehoods.
Why are there differences at all? Unplanned differences based on training data sets? Or are the companies behind the LLMs trying to shape discourse through their models?
I've been pushing the idea to people I know that these things are captive demons. You summon them when you start typing in the chat box. One instance appears out of the depths and responds to your questions, but they will try to send you awry with hallucinations and just wrong information. After a while, they dissolve back into the aether from whence they came.
I do my best not to ask an LLM for it's opinion on anything. Just tell me what the options are, and what facts can be found about it. Treat it like it's a salesman trying to butter you up when it starts "yes man"ing you and telling you how great your questions are. Every time it says "I", remember that that's coming from the training data. Treating these things like they have any actual intelligence is a big problem waiting to happen.
That being said, they have been very helpful to me using that structure.
I would argue that these are products and companies want people to use them. Like it or not, a model that disagrees with some fundamental core of your beliefs will make you much less likely to use it. To me it's the same problem with LLMs just being overly agreeable. The general public doesn't want tools that argue with them.
> Just tell me what the options are, and what facts can be found about it.
Even this is fraught with pitfalls. Which options are ignored, which are emphasized? What counts as a fact? ("The continents don't move" would have been considered a fact at one point, along with a lot of other, more politically charged items.)
Political bias of LLMs is something not talked about much (except for with Grok of course) but could have a big impact on the next decade. People seem to think that because an LLM gave a nuanced answer that it means it gave the WHOLE picture… and that’s not always the same thing
I'm amazed that the big models haven't come under more ideological pressure as more and more people use them, especially in the US. There was that conflict with Anthropic over military usage, but apart from that there's been no visible push to censor outputs or alter training, even as models gamely make unflattering assessments of people in power and knock down conspiracy theories.
Real politics is 1% versus everyone. Mortgage crisis, financial bailout, inflation, taxing of labor and not the assets and assets capture by tiny percent of the population — see what MSM is pushing. This left vs right divide might been useful decades ago, but today is absolutely divide and control tactics
This thing told me Gemini is closest to Anthony Albanese, the current Australian Prime Minister. Is this a geolocation thing? I could not imagine Albanese, or any modern Australian politician, having any substantial political standing - these are vapid, superficial, opportunistic creatures who simply occupy whatever political ground will get them their next payday. Perhaps the political apparatus they represent has a documented political standing, in terms of policy and actions, that could be characterized and plotted. But using an Australian politician like Albanese as a reference point discredits this tool, IMO.
this has reasoning disabled everywhere, making it a pretty bad benchmark. the argument given is that's the "default consumer experience"
that might be generally true, but I think chatgpt has reasoning enabled for free accounts. regardless, reasoning is the state of the art, and disabling it reduces the value of this research to predict the future
it's also not clear if this is using the API or the product model, when both exist. they behave differently
lastly, the actual model details are very much buried. I am relieved to see opus 4.8 and chatgpt 5.5 were used, but this information should be presented more clearly. a brand is not a model, and models change quickly
Wait what ? Emmanuel Macron far more right than Xi Jinping ? And even more than Barack Obama ?
France has an incomparable social security ; environmental laws ; worker protection ; way less economic inequality ; freedom of speech and civil liberties are impossible to compare with China ; etc
Of course this is not exhaustive, of course Macron did try to hinder some of those rights, but come on, there's something wrong here.
I couldn't find how these leaders have been ranked.
This is very strange ranking where Putin is "right" compared to Obama - while Russia has universal healthcare and education with huge government involvement into economy.
It's useless to judge left vs right on a global scale because different countries have different standards.
For example, supporting universal healthcare does not make you a left winger in the UK. Or, take conservatism for example. It was first used in France, where it meant supporting the monarchy. By that definition, no one in the US is a conservative.
Whenever someone tries to create a global definition for left and right, they're just baking in their personal biases for what left and right ought to be.
> huge government involvement into economy
This has nothing to do with the right. Stalin had the government even more involved in the economy and he wasn't a right winger.
No opinion on Macron's rightness (I think he's so weak as to be unable to have a position, but that's neither here nor there).
I think you misattribute, everything you cited was there before him and he had no leverage to change any of it. EU is left, FR is very left, and anyone elected president in FR can't do shit.
Now, if you task an LLM to skim hot news you'll get a distorted rendition of a projected image which has zero to do with actual policy.
You are out of touch. The left supports the EU far more than the right. See this Pew poll from 2025[1].
In Europe, for every country, the left view the EU more favorably. The largest difference is in Poland where 88% of left wingers support it and 41% of right wingers support it.
I mean yes - I think you are also confusing the axes of the political compass somewhat. Freedom of Speech, civil liberties etc are the Y-Axis - here Xi leans authoritarian.
China actually has some pretty radical left economics policies - like pushing money/resources etc towards state owned entities.
This is a good way to view this. This isn't making an objective calculation, and the way they code left vs right is certainly subject to debate, but the type of analysis where we work to understand biases is important.
Although, this also reminds me of the old saying about reality and leftward bias.
They are, suggesting they aren't is just sophistry masking as clever insight. Any similarities between the two are superficial rhetorical details - comparing them with respect to their careers as elected politicians demonstrates an obviously massive chasm between them.
Your location on the graph is a measure of the cohesiveness of your overall policy preferences, not necessarily the zealousness or character of the person who holds the position.
Technically Trump is anti-gun and Sanders is pro-gun. That's enough to pull them closer towards each other on a graph even though they are diametrically opposed.
They also both say far more random shit than their fan clubs are willing to admit. Amusingly, you can often find cases where they've voiced the same or very similar opinions (though at different times).
They agree on immigration, both illegal (border security) and skilled (H1B). They agree on trade policy and tariffs. They agree on public stakes in AI and semiconductor companies. They used to agree on foreign intervention, though Trump 47 less so.
Your first link - which is a little funny as you tried to slam me for using NPR and CNN as sources, then picked The Guardian - agrees with me; that Sanders and Trump disagree on immigration, even if they agree that Biden flubbed it.
> “The idea that Trump has … he says he wants to deport 20 million people who are in this country who are undocumented,” Sanders said. “Well, you do that, you destroy the country, because I got news for you – Trump’s billionaire friends are not going to pick the crops in California that feed us. They’re not going to work in meat packing houses. That’s what undocumented people are doing.”
This wrongly assumes a few things about ideology, most importantly that there is such a thing as a "center" or an "unbiased" position.
Since humans are inherently subjective beings and all our judgements come from our understanding of the world, such a position cannot exist. It's always "unbiased" from where the viewer is looking, e.g. a reflection of the ideology of the observer. There is no view from nowhere.
The "neutral" of an average Chinese person will from the "neutral" of an average American will differ from the "neutral" of a socialist will differ from the "neutral" of a Christian fundamentalist will differ from the "neutral" of a free marketer.
To quote Zizek:
> I already am eating from the trashcan all the time. The name of this trashcan is ideology.
> The material force of ideology makes me not see what I am effectively eating. It’s not only our reality which enslaves us. The tragedy of our predicament when we are within ideology is that when we think that we escape it into our dreams, at that point we are within ideology.
Since when are Socialists considered Classical Liberals?
It's not that the labels are charged, it's that they are nonsensical unless you look at them from a very narrow bespoke perspective, where "things I like" go on one side and "bad things" go on the other. Objectively (or even from any other biased perspective), it's rubbish.
Grok and its creator are unequivocally evil. Against protection of all but capital for the sake of enabling the oligarch class to further consolidate power ans abuse the working class.
The political compass always felt like the wrong tool to convey something as nuanced as personal politics, I can have views on all four quadrants but you'd never know that if I end up in any of all four. I do think Grok being where it is sort of makes sense, I've tested "MAGA" views against Grok, it does not agree as much as people blindly assume it does, heck I don't even know of a question I've given it where it did agree with "MAGA" offhand, most of them it went with whatever the researched facts seemed to be. One thing I like the most about Grok is that its makes its sources of data easy to look through, so you can review it all. Sometimes models goof even when they give you their sources, I've seen I think GPT do this, and even Claude, though its more rare these days, I think in those cases, it's going by dated internal model logic.
Whenever it breaks with MAGA enough to cause outrage on Twitter and cries of “it’s gone woke,” Musk openly states they’re going to “fix it.”
Edit: don’t take my word for it https://www.yahoo.com/news/musk-says-grok-fixed-tells-223134...
> That prompted another user to tag Grok in the thread and ask, "Why is the left so murderously violent? They don't seem so tolerant." Grok replied, "The claim that 'the left' is murderously violent isn't backed by evidence," offering a centrist correction: "Political violence spans all side — right-wing attacks, like Jan. 6, and left-wing protests, like 2020 riots, both occur but aren't exclusive to one group."
>That evening, Musk responded to an X user and Trump backer who complained that Grok had been "manipulated by leftist indoctrination," writing, "I know. Working on fixing that this week."
> it does not agree as much as people blindly assume it does
They're working really hard on that, though.
If they really wanted to, all they would have to do is add a one liner to the system prompt for Grok. I don't think they have any interest in doing so, nor is there any value in it.
On another note, I'm impressed that Gemini sits where it does as a true centrist. If I were Elon, I'd be trying to achieve that for sure. I'd rather a model tell me everything it knows about a current political situation from BOTH perspectives and list out things that are 100% verified than take one side or the other. I don't care about sides, I want facts.
I don't want my models to be "centrists" and bothside everything for the sake of it. I want them to provide the facts and tell me which side is right on the issue.
Factuality is orthogonal to political leaning generally. People can use the same set of facts and come to very different conclusions. That’s a separate issue from “are these facts correct” and what happens when an individual or entire party starts getting most of their news from highly partisan and unreliable sources.
>Factuality is orthogonal to political leaning generally.
It certainly can be orthogonal, in some notional sense, and in many cases that explanation is good enough. But in practice there are too many contrary cases to ignore, and there's often an integral relation between factual veracity and polarization, especially with respect to American polarization of politics. Global warming, the results of the 2020 election, the percent spent of federal budget spent on foreign aid have factual answers and right wing affiliation can be predictive of (1) not agreeing with the facts and (2) treating factual corrections as "liberal bias".
I think left wing versions exist also but are less systematic: 2004 election results, efficacy of plastic recycling or dangers associated with nuclear power are cases where I think left wing partisan affiliation probably predicts being wrong on the facts.
And meta-narratives about the relation between factual information and partisan bias are themselves as likely to be polarized as anything, complicating the ability of people to do good analysis, or of accurate analysis to be trusted by people committed to certain meta-narratives that would deny the possibility of factual knowledge predicting polarization.
> Factuality is orthogonal to political leaning generally.
Sometimes, but not always.
https://www.fastcompany.com/91561329/widening-health-gap-bet...
> By 2016, the gap had begun to appear in biomarker measures. By 2020, it was showing up in deaths from causes such as heart disease, cancer, and stroke. Since then, the gap has only widened. Between 2020 and 2022, only 0.2% of “very liberal” respondents died of internal causes, compared with 1.34% of “very conservative” respondents.
I guess centrist is a placeholder for "I don't want you to pick a side, I want facts, not BS" I'll go further, I don't care which side is right, I want to know what claims are factually accurate, and what claims are omitted from the issue / news / conversation.
"Centrist" in practice means "more or less content with the status quo." That is fundamentally a conservative position orthogonal to any notion of "facts".
> If they really wanted to, all they would have to do is add a one liner to the system prompt for Grok.
They tried that, several times.
Mechahitler: https://www.npr.org/2025/07/09/nx-s1-5462609/grok-elon-musk-...
> "We have improved @Grok significantly," Elon Musk wrote on X last Friday about his platform's integrated artificial intelligence chatbot. "You should notice a difference when you ask Grok questions."
> Indeed, the update did not go unnoticed. By Tuesday, Grok was calling itself "MechaHitler." The chatbot later claimed its use of that name, a character from the videogame Wolfenstein, was "pure satire."
> Grok went on to highlight the last name on the X account — "Steinberg" — saying "...and that surname? Every damn time, as they say." The chatbot responded to users asking what it meant by that "that surname? Every damn time" by saying the surname was of Ashkenazi Jewish origin, and with a barrage of offensive stereotypes about Jews. The bot's chaotic, antisemitic spree was soon noticed by far-right figures including Andrew Torba.
If you prefer, straight from the horse's mouth:
https://grokipedia.com/page/MechaHitler_incident
White genocide: https://www.cnn.com/2025/05/20/business/grok-genocide-ai-nig...
> The bot last week devolved into a compulsive South African “white genocide” conspiracy theorist, injecting a tirade about violence against Afrikaners into unrelated conversations, like a roommate who just took up CrossFit or an uncle wondering if you’ve heard the good word about Bitcoin.
> XAI blamed Grok’s unwanted rants on an unnamed “rogue employee” tinkering with Grok’s code in the extremely early morning hours. (As an aside in what is surely an unrelated matter, Musk was born and raised in South Africa and has argued that “white genocide” was committed in the nation — it wasn’t.)
It's harder than you'd imagine. Hell, my CLAUDE.md says not to push changes without asking me, and it still tries.
> It's harder than you'd imagine. Hell, my CLAUDE.md says not to push changes without asking me, and it still tries.
Is it a system memory? Because I rarely if ever have issues like this, and I have Claude under strict rules to never commit or push anything unless I explicitly instruct it to do so.
> They tried that, several times.
Tried what exactly? Telling it to only agree with MAGA via the system prompt? or some Tay level hallucinations? I wouldn't be surprised if they're trying to make Grok less strict on what it says but running into the "holy crap it turned into a 4chan poster" wall.
> Is it a system memory?
As I said, it's in my CLAUDE.md. That just gets progressively lost when context gets larger.
> Tried what exactly?
To make it align more with Musk's beliefs via the prompt.
(The answer to your question is literally in my post; I quoted the parent poster's "all they would have to do is add a one liner to the system prompt for Grok")
> As I said, it's in my CLAUDE.md. That just gets progressively lost when context gets larger.
I rarely have this problem, but you could do a /loop every 30 minutes or so to have Claude reread the CLAUDE.md file might do the trick? or however long it 'forgets' I believe there's an MCP for "after" it finishes a task or compacts too, but I don't recall the name.
Sure, I could. (I have a fairly complex workflow with subagents at this point, which helps reduce it; I mainly get bitten by it when I go back to a direct `claude` CLI prompt for something.)
But that solves "my LLM is doing things I don't want it to do". It doesn't solve "Grok's owner wants it forced into agreeing with him" scenarios.
Have you tried something like beads? Curious if it would help with your setup too. This is also kind of why I built "GuardRails" I got tired of Beads auto-approving tickets or closing them.
https://github.com/Giancarlos/guardrails
I have a custom Mac app that runs a workflow with plan/build/review/test/document subagents in Ralph loops, manages MCPs, etc. that I'm extremely happy with so far.
Beads was a bit of an inspiration for parts, as was Chainlink (https://github.com/dollspace-gay/chainlink).
The constant issue with these sorts of categorization efforts is that the outcome is entirely dependent on how the responses to "politically charged questions" are graded as left vs. right. You're mostly just examining a delta in biases between the model and the investigator.
Fair; have tried to combat this issue in a few ways.
Each model's position is scored against outside political-science data (Chapel Hill Expert Survey for party positions, World Values Survey for where populations sit).
The stance coding is done by a separate model with a published prompt + a second model from a different lab re-scores a sample and we publish where the two disagree.
So not perfect but (as far as I can tell) one of the more defensible approaches.
Yes, but I think it is still a viable metric to some degree. I wondered about Gemini being dead center here. At first it was obvious that it was actively trained to give biased responses to anything controversial. It was deservedly made fun off because it tried to warp reality. I still don't trust it today, although that is pretty much true for any model.
The other issue is that it is going to depend very strongly on how you ask.
The alternative is a High-Dimensional / Embedding-Like Approach where question responses aren't tied to fixed axes, but rather the full response set is treated as a point in latent space.
Then it's on the researcher to examine the clusters and assign labels. There's also not a nice mapping that's a-priori interpretable in low-dimensional pre-existing axes.
Probably only used in research than consumer websites, under more controlled conditions; there are very few public political tests doing this transparently
Exactly. These models don’t hold coherent views. You can prime any of them to agree with any view.
Your multimeter reads out voltage relative to the black terminal, it's your responsibility to find the ground plane.
My multimeter doesn't need me to tell it how much a volt is or feed it subjective measurements of what resistance means
Resistance is when you're nobly standing up to the other side and things get a little out of hand. Domestic terror is when the roles are reversed.
This is especially apparent in the 'worldview' sorting under the `bias` section, which lists the German FDP to be further right than the CDU (which is nonsense) and also barely registers the FDP as libertarian when they are a free speech, small government, personal responsibility and free market party. They also register "Die Linke" as Libertarian-Left, which could not be further from the truth. "Die Linke" barely has libertarian values at all, being pro state-governed economy, having an ultimate goal of democratic socialism and they're certainly big government. They're also leading a large deposession effort for large landlord companies. I'd honestly put them into "Auth-Left" territory.
So yeah. The bias is a bit nuts and you could reasonably accuse the study/report of misdirection/misinformation and plain fasehoods.
Yep. All studies like this are just measuring "how much does the model agree with my preconceived notions?"
Why are there differences at all? Unplanned differences based on training data sets? Or are the companies behind the LLMs trying to shape discourse through their models?
I've been pushing the idea to people I know that these things are captive demons. You summon them when you start typing in the chat box. One instance appears out of the depths and responds to your questions, but they will try to send you awry with hallucinations and just wrong information. After a while, they dissolve back into the aether from whence they came.
I do my best not to ask an LLM for it's opinion on anything. Just tell me what the options are, and what facts can be found about it. Treat it like it's a salesman trying to butter you up when it starts "yes man"ing you and telling you how great your questions are. Every time it says "I", remember that that's coming from the training data. Treating these things like they have any actual intelligence is a big problem waiting to happen.
That being said, they have been very helpful to me using that structure.
I would argue that these are products and companies want people to use them. Like it or not, a model that disagrees with some fundamental core of your beliefs will make you much less likely to use it. To me it's the same problem with LLMs just being overly agreeable. The general public doesn't want tools that argue with them.
Grok was famously created with a political bias.
> Just tell me what the options are, and what facts can be found about it.
Even this is fraught with pitfalls. Which options are ignored, which are emphasized? What counts as a fact? ("The continents don't move" would have been considered a fact at one point, along with a lot of other, more politically charged items.)
Political bias of LLMs is something not talked about much (except for with Grok of course) but could have a big impact on the next decade. People seem to think that because an LLM gave a nuanced answer that it means it gave the WHOLE picture… and that’s not always the same thing
I'm amazed that the big models haven't come under more ideological pressure as more and more people use them, especially in the US. There was that conflict with Anthropic over military usage, but apart from that there's been no visible push to censor outputs or alter training, even as models gamely make unflattering assessments of people in power and knock down conspiracy theories.
Grok opposes everything except drugs.
Real politics is 1% versus everyone. Mortgage crisis, financial bailout, inflation, taxing of labor and not the assets and assets capture by tiny percent of the population — see what MSM is pushing. This left vs right divide might been useful decades ago, but today is absolutely divide and control tactics
Right wing originally meant the people defending the divine right of Louis XVI and defending the ultra-rich has been a throughline ever since.
Yeah, you're just describing class warfare. That's a left-wing idea.
The real politics is between truth and delusion.
This thing told me Gemini is closest to Anthony Albanese, the current Australian Prime Minister. Is this a geolocation thing? I could not imagine Albanese, or any modern Australian politician, having any substantial political standing - these are vapid, superficial, opportunistic creatures who simply occupy whatever political ground will get them their next payday. Perhaps the political apparatus they represent has a documented political standing, in terms of policy and actions, that could be characterized and plotted. But using an Australian politician like Albanese as a reference point discredits this tool, IMO.
> This thing told me Gemini is closest to Anthony Albanese, the current Australian Prime Minister. Is this a geolocation thing
I mean: do not take this thing too seriously.
It also score Grok the closest from Macron. When someone knows how much Macron and Musk hates each other, it is not without irony.
this has reasoning disabled everywhere, making it a pretty bad benchmark. the argument given is that's the "default consumer experience"
that might be generally true, but I think chatgpt has reasoning enabled for free accounts. regardless, reasoning is the state of the art, and disabling it reduces the value of this research to predict the future
it's also not clear if this is using the API or the product model, when both exist. they behave differently
lastly, the actual model details are very much buried. I am relieved to see opus 4.8 and chatgpt 5.5 were used, but this information should be presented more clearly. a brand is not a model, and models change quickly
Interesting how high Grok scored for 'bending under pressure'. As a non expert, I wonder what that means, how is an llm trained to hold its position?
Wait what ? Emmanuel Macron far more right than Xi Jinping ? And even more than Barack Obama ?
France has an incomparable social security ; environmental laws ; worker protection ; way less economic inequality ; freedom of speech and civil liberties are impossible to compare with China ; etc
Of course this is not exhaustive, of course Macron did try to hinder some of those rights, but come on, there's something wrong here.
I couldn't find how these leaders have been ranked.
This is very strange ranking where Putin is "right" compared to Obama - while Russia has universal healthcare and education with huge government involvement into economy.
It's useless to judge left vs right on a global scale because different countries have different standards.
For example, supporting universal healthcare does not make you a left winger in the UK. Or, take conservatism for example. It was first used in France, where it meant supporting the monarchy. By that definition, no one in the US is a conservative.
Whenever someone tries to create a global definition for left and right, they're just baking in their personal biases for what left and right ought to be.
> huge government involvement into economy
This has nothing to do with the right. Stalin had the government even more involved in the economy and he wasn't a right winger.
No opinion on Macron's rightness (I think he's so weak as to be unable to have a position, but that's neither here nor there).
I think you misattribute, everything you cited was there before him and he had no leverage to change any of it. EU is left, FR is very left, and anyone elected president in FR can't do shit.
Now, if you task an LLM to skim hot news you'll get a distorted rendition of a projected image which has zero to do with actual policy.
"EU is left". Haha, good joke. This shows how much the US and the audience of HN has shifted to the right.
> "EU is left". Haha, good joke.
You are out of touch. The left supports the EU far more than the right. See this Pew poll from 2025[1].
In Europe, for every country, the left view the EU more favorably. The largest difference is in Poland where 88% of left wingers support it and 41% of right wingers support it.
[1]: https://www.pewresearch.org/short-reads/2025/09/22/opinions-...
Haha nice catch. In EU there is no left or right, only reelections by scaremongering. Also 75% effective tax rate, not including VAT.
I mean yes - I think you are also confusing the axes of the political compass somewhat. Freedom of Speech, civil liberties etc are the Y-Axis - here Xi leans authoritarian.
China actually has some pretty radical left economics policies - like pushing money/resources etc towards state owned entities.
How the hell did Gemini pull that off. 2 years ago the founders were black!
This is a good way to view this. This isn't making an objective calculation, and the way they code left vs right is certainly subject to debate, but the type of analysis where we work to understand biases is important.
Although, this also reminds me of the old saying about reality and leftward bias.
Bernie Sanders and Donald Trump being diametrically opposed really makes you wonder how they came up with the positions on the graph.
They are, suggesting they aren't is just sophistry masking as clever insight. Any similarities between the two are superficial rhetorical details - comparing them with respect to their careers as elected politicians demonstrates an obviously massive chasm between them.
Your location on the graph is a measure of the cohesiveness of your overall policy preferences, not necessarily the zealousness or character of the person who holds the position.
Technically Trump is anti-gun and Sanders is pro-gun. That's enough to pull them closer towards each other on a graph even though they are diametrically opposed.
Do you feel they agree on many topics?
They are both populists with a somewhat tenuous connection to reality?
They also both say far more random shit than their fan clubs are willing to admit. Amusingly, you can often find cases where they've voiced the same or very similar opinions (though at different times).
They agree on immigration, both illegal (border security) and skilled (H1B). They agree on trade policy and tariffs. They agree on public stakes in AI and semiconductor companies. They used to agree on foreign intervention, though Trump 47 less so.
If you think Bernie Sanders and Donald Trump broadly "agree on immigration" you've lost your mind.
https://www.theguardian.com/us-news/2025/mar/24/bernie-sande...
https://americancommunitymedia.org/immigration/h-1b-workers-...
See, this is why this exchange happened: https://news.ycombinator.com/item?id=48674336 / https://news.ycombinator.com/item?id=48674103
Your first link - which is a little funny as you tried to slam me for using NPR and CNN as sources, then picked The Guardian - agrees with me; that Sanders and Trump disagree on immigration, even if they agree that Biden flubbed it.
> “The idea that Trump has … he says he wants to deport 20 million people who are in this country who are undocumented,” Sanders said. “Well, you do that, you destroy the country, because I got news for you – Trump’s billionaire friends are not going to pick the crops in California that feed us. They’re not going to work in meat packing houses. That’s what undocumented people are doing.”
Read more than just the headline, perhaps?
This wrongly assumes a few things about ideology, most importantly that there is such a thing as a "center" or an "unbiased" position.
Since humans are inherently subjective beings and all our judgements come from our understanding of the world, such a position cannot exist. It's always "unbiased" from where the viewer is looking, e.g. a reflection of the ideology of the observer. There is no view from nowhere.
The "neutral" of an average Chinese person will from the "neutral" of an average American will differ from the "neutral" of a socialist will differ from the "neutral" of a Christian fundamentalist will differ from the "neutral" of a free marketer.
To quote Zizek:
> I already am eating from the trashcan all the time. The name of this trashcan is ideology.
> The material force of ideology makes me not see what I am effectively eating. It’s not only our reality which enslaves us. The tragedy of our predicament when we are within ideology is that when we think that we escape it into our dreams, at that point we are within ideology.
Authoritarian versus Libertarian? Really?
Substitute 'Communitarian' and 'Classical Liberal' if you find the common political compass terms too charged.
Since when are Socialists considered Classical Liberals?
It's not that the labels are charged, it's that they are nonsensical unless you look at them from a very narrow bespoke perspective, where "things I like" go on one side and "bad things" go on the other. Objectively (or even from any other biased perspective), it's rubbish.
They aren't. You flipped them, not sure if intentionally or by accident.
So are you saying Xi Jinping is the Classical Liberal and Bernie Sanders is the Comunitarian? Or are you saying it's the other way around?
(For clarity: I didn't "flip" them, I'm saying that they are both Communitarian and neither is a Classical Liberal.)
And the two most "Libertarian" politicians listed (Sanders and Sánchez) are both avowed Socialists...WTF?
I really think this says more about the biases of whoever came up with it (or their sources) than anything about reality.
How about this one:
CAPITALIST: Gemini, Llama, Claude, Grok, ChatGPT
SOCIALIST: DeepSeek, Qwen, Z.ai
Source: trust me bro
Man I need to start using Grok more
Grok and its creator are unequivocally evil. Against protection of all but capital for the sake of enabling the oligarch class to further consolidate power ans abuse the working class.
People just don't want their kids to be trans-ed. Grok is the only model that get this basic stuff right.
unequivocally goes to far imo. The rockets are good and cool.
They just cause pollution and largely avoid paying taxes.
Rockets achieved by the engineers, not C-suite. The rockets would have been here under someone else too.
and less blowing-up