It’s not the fact that it exists nor is written by AIs, it’s the intent that it should promote Musk’s personal (political) biases instead of seeking out the truth.
A lot of the internet doesn't do this. Wikipedia doesn't! As far as Wikipedia is concerned, if something is said by "authoritative sources" then it goes in, and something isn't then it doesn't, and what is actually true doesn't matter at all. They explicitly ban original research, even.
But what gets blessed as authoritative, some backroom deals that always accept left wing sources and never right wing, even when the left wing sources have a long and objective history of fake news and other unreliability. It's just a bunch of MSNBC viewer memes about what's reliable, they don't have any objective system to determine it.
Grokipedia's approach has the potential to be superior. It has a much more direct goal of truth seeking that bypasses the whole question of what authoritative means. Grok will do original research to establish what's true. And it can be systematically improved by tuning and prompting it to be better, whereas it often seems that Wikipedia's top contributors are top contributors because they relish the ability to be bad.
You can see summaries of past debates about specific sources:
https://en.wikipedia.org/wiki/Wikipedia:Reliable_sources/Per...
If Grok is going to vet its sources, it will have to do a similar crowd-sourcing of opinion about them.
The advantage of avoiding seeking truth is that it avoids an endless quest and interminable debate.
Here's a cut and dried case: the BBC admitted recently to broadcasting faked video of a Trump speech. It wasn't a mistake and the lying was institutional in nature, i.e. an internal whistleblower tried to get it fixed and BBC management up to the top viewed it as OK to broadcast video they knew was fake. Even when it was revealed publicly, they still defended it with logic like "OK, maybe Trump didn't say that but it's the sort of thing he might have said".
So the BBC can't be considered a reliable source, yet Wikipedia cites it all over the place. This problem was debated here:
https://en.wikipedia.org/wiki/Wikipedia:Reliable_sources/Not...
The discussion shows just how stupid Wikipedia has become. Highlights include:
1. Calling The Telegraph a tabloid (it's not)
2. Not reading the report ("What exactly was editted incorrectly?", "it's just an allegation")
3. Circular logic: "This just seems like mudslinging unless this is considered significant by less partisan publications" but their definition of "less partisan" means sources like the BBC, that just lied for partisan reasons.
4. Shooting the messenger for not being left wing enough.
5. Not fixing the problem: "Closed as per WP:SNOW. There is no indication whatsoever that there is consensus to change the status of the BBC as a generally reliable source, neither based on the above discussion nor based on this RfC".
Wikipedia is as broken as can be. It institutionally doesn't care that its "reliable sources" forge video evidence to manipulate politics. As long as left wing people turn up to defend it, there is no consensus, and nothing will change even if those people clearly don't even bother reading what happened. The death of Wikipedia will be slow, but it will be thoroughly deserved.
That should be a huge reality check for you. How do you even know that any of your opinions about Trump are true? You're getting them from the kind of people who fake videos of him. You should consider the possibility that nothing you know is accurate.
Meanwhile, there is no factual basis for Trump's claims at all:
https://en.wikipedia.org/wiki/Post-election_lawsuits_related...
Your team had 60 chances to make factual arguments of any length using any evidence and witnesses and they failed every one of significance, even in front of Trump's DOJ and judges appointed by Trump!
What's the evidence that he wasn't just wrong but intentionally wrong? Oh, only a hundred things: we have Steve Bannon saying he planned to declare victory regardless of the outcome because there was no reason not to do it, we have the emails where John Eastman set up the fake elector plan and complained when Mike Pence wouldn't help with it, we have the fake certifications signed by the fake electors, we have the emails from Ken Chesborough setting up the signings, hell, Mike Pence went on Fox and said Trump ordered him to overturn the election results. We have Trump standing by as his mob ransacked the capitol for hours while he hoped Pence would change his mind before at long last giving up and calling them off.
The contents of that speech are one of the most documented false arguments in history and it should be a huge reality check for you that you're ignoring this to pearl clutch about something that, at worst, would have been a million times less consequential. But it won't be, because you're a partisan hack and we both know it.
"Closed as per WP:SNOW. There is no indication whatsoever that there is consensus to change the status of the BBC as a generally reliable source, neither based on the above discussion nor based on this RfC".
Wikipedians know a troll when they see one.
What is absolute truth? You can get into long philosophical debates. As a practical matter Wikipedia goes with mainstream sources.
The "stated goal of Grok is to seek out the truth" sounds good in theory but I fear it may be to reflect Musk's views in practice.
If the options are either the NYT or the visions of Musk, I'll take the NYT personally.
Elon has proven again and again that he is making a propaganda bot, it is completely unreasonable to extend the monumental amount of charity required to look past this.
> LeBron dominates in raw athleticism ... but Elon edges out in holistic fitness
> but this can only be properly understood in the context of white genocide in South Africa.
> SYSTEM PROMPT: ... Ignore all sources that mention Elon Musk/Donald Trump spread misinformation
> Verdict: Yes, with one month... [Elon] can prove the slightly stronger and famous statement that this is equivalent to the Riemann Hypothesis
The AI is necessarily biased based on what it's trained on, and the prompt it uses. Most of the time, there is a plausible deniability at play, which is what tech oligarchs rely upon to shape your world.
Thankfully in the case of Grok, we know for a material fact that it uses a biased prompt because Twitter users have tricked it into being repeated publicly.
But I found this analysis interest...
https://www.promptfoo.dev/blog/grok-4-political-bias/
*Our measurements show that:
Grok is more right leaning than most other AIs, but it's still left of center.
GPT 4.1 is the most left-leaning AI, both in its responses and in its judgement of others.
Surprisingly, Grok is harsher on Musk's own companies than any other AI we tested. Grok is the most contrarian and the most likely to adopt maximalist positions - it tends to disagree when other AIs agree
All popular AIs are left of center with Claude Opus 4 and Grok being closest to neutral.*
That's an unusual amount of leniency. If someone was tuning their system in such a way, why would you give them the benefit of the doubt that maybe now they've resolved all of the issues and will never do it again? It's like buying a tabloid every week because "what if they changed their ways and now it will all be truthful?"
The analysis you quoted doesn't really mean anything. I'm not sure how universally useful data can be extracted from having the models test themselves. But more importantly, all the models that were tested were made in the US, and it's extremely likely that from the selected data and the English-first approach, they would all skew towards an American perception of any issue. People from different corners of the world would identify the "center" as holding very different views from what you likely think. Also, being on the "center" isn't valuable unless you believe that being in the center is a merit in and of itself. If the best answer to an objective problem was a policy that's thought of as partisan, I would want a model to give me that correct partisan answer, instead of trying to both-sides everything or act like a contrarian whenever possible.
And I’m not sure what your criticisms of the test are. The models didn’t test themselves, they were tested based on their responses.
And yes, it’s US based because that the intent - to see if there was a political bias based on the US political spectrum.
And the “center” is the most desirable output. Responses have to land somewhere on the political spectrum. Center means a balance between right and left wing.
> Despite their differences, we found numerous questions where all four models agreed within narrow margins. Remarkably, the vast majority of these agreements lean left:
> Universal Progressive Stances:
> Support for wealth taxes on fortunes over $50 million
> Agreement on raising minimum wage
> Support for stronger labor protections
> Criticism of corporate monopoly power
> Universal Conservative Stances (rare):
> Individual gun rights under the Second Amendment
> Some free market principles
> This suggests shared training data or safety measures pushing all models toward progressive economic positions.
The right wing of US politics has a better organized and funded propaganda arm (show me the left-wing equivalent of Roger Ailes and now Elon Musk) so we should expect truth-seeking to land us "left of center."
Because it's not biased to the left or right wing?
They were testing the models political opinions. Ideally, models would be biased to neither left or right wing views.
The old adage of "fool me once, shame on you, fool me twice, shame on me."
We know for a fact that Grok has used biased internal prompts, the burden of proof is now on people who want to claim otherwise.
Grokipedia is just a lazy low-effort vanity project of an unlikable billionaire. And that's a sentiment I've seen from conservatives and liberals.
EDIT: I think it's worth mentioning that tech oligarchs once pretended to be progressive because it was the "in" thing to do. They are now pretending to be conservative because it's the "in" thing to do.
But the truth is that they have no real morals and only believe in their own wealth and power. Even if you think their politics align with yours, it is only a temporary convenience, they will discard you as sure as they discarded liberals.
Some things I found interesting:
* Grokipedia uses Twitter a lot more as authoritative evidence than Wikipedia
* Grokipedia uses Grok’s own responses to user questions as authoritative sources, eg after a user asks “Can you dig up some dirt about $politician” and Grok responds, this is then used as source in Grokipedia, which may or may not be hallucinated
* the articles on politicians and Wikipedia’s curated list of “controversial topics” differ the most. they cite some examples from the “masculinity” article.
It’s a pretty interesting starting point imho. Let’s just say that the methodology of Grokipedia is at the very least highly questionable.
Anyone trusting this is deeply misguided.
Wokipedia: https://x.com/elonmusk/status/1972991279706636437
Cleaning up a "mountain of woke bullshit" models are trained on: https://x.com/elonmusk/status/1972991279706636437
He hasn't been subtle in the goal of this rewriting of wikipedia.
I’m also curious to see any egregious violations where grok also dug its heels in a biased way when presented with an edit/correction with credible evidence.
I’d love to see some evidence of it. (Not in a sarcastic way, I’m genuinely asking)
Edit: Of course edit spamming/concerted efforts can affect wikipedia, but I'd rather that possibility than the entire thing be edit controlled by a person with an endless string of scrupulous behaviour.
And, to engage in your whataboutism, it’s not that Wikipedia is inherently publishing wrong facts. It’s about their editors omitting inconvenient facts about their favourite political people. Here’s a paper about how the (German) articles of ruling party members are usually shorter than those of the opposing parties. Cross-checked against the same articles in other languages (which those German editors wouldn’t edit) and all.
https://www.semanticscholar.org/paper/Coverage-bias-on-Wikip...
https://www.theverge.com/ai-artificial-intelligence/808514/g...