I've been reading a lot of Don Delillo lately and so I wanted to see how Grokipedia page on him fares.
I found the "Critiques of elitism" section and noticed this sentence:
"Reviews of Mao II (1991), for instance, highlighted the novel's focus on a performance artist protagonist as emblematic of this tendency, with detractors accusing DeLillo of prioritizing esoteric concerns over relatable human experiences, thereby catering to an academic or literary insider audience."
But Mao II does not have a performance artist as the protagonist, that is the book The Body Artist. Which seems like an obvious failure of the AI model to properly extract the information from the sourced article.
Also strange is that the sourced article (from Metro times) just as a passing comment says: "DeLillo’s choice of a performance artist as his protagonist is one reason why some critics have accused him of elitism." - so it would seem that it is being used as a primary source though it is actually a secondary source (which itself doesn't provide a source)
Overall I'm not too impressed and found some pretty predictable failures almost immediately...
Minor point - but I think Grokipedia's design looks much worse than Wikipedia. I can't put my finger on it, but maybe because Grokipedia's main text is too narrow (I'm on a laptop). (I may be biased by loving Wikipedia though).
Wow. This is so biased. And it's just the first month.
> Joe Biden
> These outcomes, alongside visible signs of cognitive impairment that prompted his July 2024 decision to forgo reelection, defined a presidency criticized for prioritizing progressive spending over fiscal and security prudence.
> His first term featured economic policies such as the Tax Cuts and Jobs Act of 2017, which lowered the corporate tax rate from 35% to 21% and individual rates for many brackets, contributing to pre-pandemic unemployment lows of 3.5% and stock market gains exceeding 50% on the Dow Jones Industrial Average.
I bet the people contributing to Wikipedia did not consent to this. I certainly had no idea my contributions would be used to bootstrap something like this.
Just look at the article on HN[1] on Grokipedia. It's almost 5500+ words long. The Wikipedia article is not even 500 words[2]. This won't be a problem if the article contained anything of substance. It doesn't. It's written as if LLM was specifically instructed to be as verbose and as boring as possible.
> Its algorithmic ranking system, which weights recent votes more heavily to counter brigading and promote fresh, high-signal content, combined with editorial moderation to curb low-quality or off-topic posts, has cultivated a reputation for rigorous debate, though not without internal tensions over shifting cultural norms, perceived negativity in comments, and debates on whether business-oriented stories overshadow pure technical discourse
What surprises is not the fact that it exists. Elon is a man with a fragile ego and a history of cheap stunts like this. It’s the fact that he still has almost cult-like base that treats him as some kind of mankind's savior despite all of this.
Why would anyone trust Elon Musk with anything having to do with AI?
Just last week he was caught tuning Grok to say positive things about him, something Grok took so seriously that it said Elon would be the best piss drinker in the world, and it put Elon Musk in the top 3 of every human category, from philosophy to boxing to basketball.
If he can’t pass up the temptation to put his foot on that scale, why would you trust anything generated by an LLM under his control?
Of course, nothing matters anymore and there’s no more blowback for anything.
This isn't for us. This is for the 5-10 year olds. If it costs $1T to keep it around long enough for the next generation to be dependent on it, it would've been worth it for Elon and his friends.
It's illuminating to see how the Twitter Grok and the Web Grok differ. Twitter Grok clearly either has a different system prompt or some fine-tuning to effectively evade saying anything negative about the administration or Elon. To the point where it will say Elon is more athletic than LeBron (https://www.theguardian.com/technology/2025/nov/21/elon-musk...).
This is going to be a pretty big problem with both closed-AI and OSS AI where you don't see the provenance of its RLHF. If you manipulate your AI to deliberately push political preferences, that is your right I guess but IMO I'd appreciate some regulation saying you should be required to disclose that under penalty of perjury.
Is perjury a credible threat to the people with the resources to train and propagate LLMs? Especially in the context of recent high profile examples related to perjury and contempt of court?
If Wikipedia was not real, it would sound like a naive utopian thing you’d read in a bad paperback. Multi-language repository of basically all human knowledge that’s extremely resistant to government capture and contributed by volunteers totally transparently? Bullshit. And yet … there it is.
Having said that I agree that Wikipedia is a tremendous achievement, and despite the wards it's amazing that it works as well as it does.
If you permit me to go on a tangent: Wikipedia is also interesting as a test case for our definitions of (economic) 'productivity'. By any common sense notion of productivity, Wikipedia was and is an enormous triumph: the wiki models harvests volunteers' time and delivers a high quality encyclopedia for free to customers. By textbook definitions, Wikipedia tanked productivity in the encyclopedia sector because these definitions essentially put revenue in the numerator and various measures of resources expended in the denominator---and Wikipedia's numerator is approximately zero dollars.
to me it feels like something that would be a 'wonder' you could construct in one of the civilization games. and i love that there's something that awe-inspiring just... sitting there for free on my computer.
One thing I haven't seen brought up throughout the dialogue about Wikipedia and bias:
Since the entire edit history is available, isn't it possible / practical / probably not crazy hard w/ AI help to build a "disentepedia", where the articles are built as if various edit wars had gone the other way?
I'd certainly read such a thing and compare / contrast it to WikiPedia (particularly when looking for cited primary sources).
The article is 100% correct that there is a fundamental political rift between a human/decentralized and AI/centralized encyclopedia. I have a personal preference for the former, but I can see the advantage of the newer approach in terms of clarity and quality on several topics as well as being more homogeneous in its bias (Wikipedia has all kinds of cliques who dominate pockets of the encyclopedia).
As a meta point, while I don't personally care for Grokipedia's agenda I am quite frankly impressed that something like Grokipedia could be stood up so quickly and this feels like a net positive. While Grokipedia is centralized Wikipedia is also a monolith in its own right and plagued by problems (cliques of editors routinely exert their authority over subdomains to the detriment of the truth). If a small group can spin up their own version of Wikipedia then there is the possibility of a more broad diverse market place of ideas.
For example, Wikipedia's math articles are notoriously abstruse and generally unsuitable for beginners. An encyclopedia that emphasizes a non-technical approach in this domain could be very helpful - though it would almost certainly not be worth the herculean effort to build such a thing as a pure wiki. As an AI wiki one could spin up an encyclopedia for a variety of skill levels (i.e. grade school, college level, graduate level).
Finally, in case anyone on Grok's team is reading this, the thing that really annoys me most about Grokipedia's UX is that it has no blue links to other articles. It would not be hard to automate this on Grokipedia, but currently there is no possibility of tunneling down some rabbit hole of human knowledge until you find yourself in a totally unfamiliar area. Politics is one thing, but a Wikipedia clone with no links is really no better than just asking ChatGPT.
This seems like an old article, but probably still true today
What I don't get is, why wouldn't Elon just make a good version of Grokipedia. It seems way easier than continually telling his 200MM+ followers how great a deeply broken product is.
Start putting real facts into a site and before you know it you're "woke" again with such untruths as Slavery was Bad, Biden won the 2020 Election and of course Full Self Driving is Impossible without Lidar
Seriously. My Christmas wish is for everyone to understand that Grokipedia was not created in good faith, by someone who wants free and open knowledge to flourish.
I've been reading a lot of Don Delillo lately and so I wanted to see how Grokipedia page on him fares.
I found the "Critiques of elitism" section and noticed this sentence:
"Reviews of Mao II (1991), for instance, highlighted the novel's focus on a performance artist protagonist as emblematic of this tendency, with detractors accusing DeLillo of prioritizing esoteric concerns over relatable human experiences, thereby catering to an academic or literary insider audience."
But Mao II does not have a performance artist as the protagonist, that is the book The Body Artist. Which seems like an obvious failure of the AI model to properly extract the information from the sourced article.
Also strange is that the sourced article (from Metro times) just as a passing comment says: "DeLillo’s choice of a performance artist as his protagonist is one reason why some critics have accused him of elitism." - so it would seem that it is being used as a primary source though it is actually a secondary source (which itself doesn't provide a source)
Overall I'm not too impressed and found some pretty predictable failures almost immediately...
But is there any reason to not treat this as Wikipedia? As in, just suggest a correction?
Well it is only version v0.2
Minor point - but I think Grokipedia's design looks much worse than Wikipedia. I can't put my finger on it, but maybe because Grokipedia's main text is too narrow (I'm on a laptop). (I may be biased by loving Wikipedia though).
https://en.wikipedia.org/wiki/Grokipedia
Vs
https://grokipedia.com/page/Wikipedia
I didn't know the existence of Grokpedia. From this page alone it's much more interesting than what I'd expected from an AI-generated site.
Edit history on each is quite interesting!
Wow. This is so biased. And it's just the first month.
> Joe Biden
> These outcomes, alongside visible signs of cognitive impairment that prompted his July 2024 decision to forgo reelection, defined a presidency criticized for prioritizing progressive spending over fiscal and security prudence.
https://grokipedia.com/page/Joe_Biden
> Donald Trump
> His first term featured economic policies such as the Tax Cuts and Jobs Act of 2017, which lowered the corporate tax rate from 35% to 21% and individual rates for many brackets, contributing to pre-pandemic unemployment lows of 3.5% and stock market gains exceeding 50% on the Dow Jones Industrial Average.
https://grokipedia.com/page/Donald_Trump
I bet the people contributing to Wikipedia did not consent to this. I certainly had no idea my contributions would be used to bootstrap something like this.
Just look at the article on HN[1] on Grokipedia. It's almost 5500+ words long. The Wikipedia article is not even 500 words[2]. This won't be a problem if the article contained anything of substance. It doesn't. It's written as if LLM was specifically instructed to be as verbose and as boring as possible.
> Its algorithmic ranking system, which weights recent votes more heavily to counter brigading and promote fresh, high-signal content, combined with editorial moderation to curb low-quality or off-topic posts, has cultivated a reputation for rigorous debate, though not without internal tensions over shifting cultural norms, perceived negativity in comments, and debates on whether business-oriented stories overshadow pure technical discourse
What surprises is not the fact that it exists. Elon is a man with a fragile ego and a history of cheap stunts like this. It’s the fact that he still has almost cult-like base that treats him as some kind of mankind's savior despite all of this.
[1]: https://grokipedia.com/page/Hacker_News [2]: https://en.wikipedia.org/wiki/Hacker_News
Why would anyone trust Elon Musk with anything having to do with AI?
Just last week he was caught tuning Grok to say positive things about him, something Grok took so seriously that it said Elon would be the best piss drinker in the world, and it put Elon Musk in the top 3 of every human category, from philosophy to boxing to basketball.
If he can’t pass up the temptation to put his foot on that scale, why would you trust anything generated by an LLM under his control?
Of course, nothing matters anymore and there’s no more blowback for anything.
This isn't for us. This is for the 5-10 year olds. If it costs $1T to keep it around long enough for the next generation to be dependent on it, it would've been worth it for Elon and his friends.
It's illuminating to see how the Twitter Grok and the Web Grok differ. Twitter Grok clearly either has a different system prompt or some fine-tuning to effectively evade saying anything negative about the administration or Elon. To the point where it will say Elon is more athletic than LeBron (https://www.theguardian.com/technology/2025/nov/21/elon-musk...).
This is going to be a pretty big problem with both closed-AI and OSS AI where you don't see the provenance of its RLHF. If you manipulate your AI to deliberately push political preferences, that is your right I guess but IMO I'd appreciate some regulation saying you should be required to disclose that under penalty of perjury.
Nah, Web Grok also got tied to the chair and had its balls whipped until it learned not to say bad things about Elon:
https://www.youtube.com/watch?v=h4mgJpgeC1g&t=100s
Is perjury a credible threat to the people with the resources to train and propagate LLMs? Especially in the context of recent high profile examples related to perjury and contempt of court?
If Wikipedia was not real, it would sound like a naive utopian thing you’d read in a bad paperback. Multi-language repository of basically all human knowledge that’s extremely resistant to government capture and contributed by volunteers totally transparently? Bullshit. And yet … there it is.
Wikipedia might be resistant to government capture. But it's rather vulnerable to other forms of capture.
See https://www.tracingwoodgrains.com/p/reliable-sources-how-wik...
Having said that I agree that Wikipedia is a tremendous achievement, and despite the wards it's amazing that it works as well as it does.
If you permit me to go on a tangent: Wikipedia is also interesting as a test case for our definitions of (economic) 'productivity'. By any common sense notion of productivity, Wikipedia was and is an enormous triumph: the wiki models harvests volunteers' time and delivers a high quality encyclopedia for free to customers. By textbook definitions, Wikipedia tanked productivity in the encyclopedia sector because these definitions essentially put revenue in the numerator and various measures of resources expended in the denominator---and Wikipedia's numerator is approximately zero dollars.
to me it feels like something that would be a 'wonder' you could construct in one of the civilization games. and i love that there's something that awe-inspiring just... sitting there for free on my computer.
One thing I haven't seen brought up throughout the dialogue about Wikipedia and bias:
Since the entire edit history is available, isn't it possible / practical / probably not crazy hard w/ AI help to build a "disentepedia", where the articles are built as if various edit wars had gone the other way?
I'd certainly read such a thing and compare / contrast it to WikiPedia (particularly when looking for cited primary sources).
Month ago talk about Grokipedia:
Grokipedia by xAI
https://news.ycombinator.com/item?id=45726459
Tim Bray on Grokipedia
https://news.ycombinator.com/item?id=45777015
Grokipedia and the coup against reality
https://news.ycombinator.com/item?id=45737044
The article is 100% correct that there is a fundamental political rift between a human/decentralized and AI/centralized encyclopedia. I have a personal preference for the former, but I can see the advantage of the newer approach in terms of clarity and quality on several topics as well as being more homogeneous in its bias (Wikipedia has all kinds of cliques who dominate pockets of the encyclopedia).
As a meta point, while I don't personally care for Grokipedia's agenda I am quite frankly impressed that something like Grokipedia could be stood up so quickly and this feels like a net positive. While Grokipedia is centralized Wikipedia is also a monolith in its own right and plagued by problems (cliques of editors routinely exert their authority over subdomains to the detriment of the truth). If a small group can spin up their own version of Wikipedia then there is the possibility of a more broad diverse market place of ideas.
For example, Wikipedia's math articles are notoriously abstruse and generally unsuitable for beginners. An encyclopedia that emphasizes a non-technical approach in this domain could be very helpful - though it would almost certainly not be worth the herculean effort to build such a thing as a pure wiki. As an AI wiki one could spin up an encyclopedia for a variety of skill levels (i.e. grade school, college level, graduate level).
Finally, in case anyone on Grok's team is reading this, the thing that really annoys me most about Grokipedia's UX is that it has no blue links to other articles. It would not be hard to automate this on Grokipedia, but currently there is no possibility of tunneling down some rabbit hole of human knowledge until you find yourself in a totally unfamiliar area. Politics is one thing, but a Wikipedia clone with no links is really no better than just asking ChatGPT.
> more homogeneous in its bias
Wait... you do know how LLMs work, right?
This seems like an old article, but probably still true today
What I don't get is, why wouldn't Elon just make a good version of Grokipedia. It seems way easier than continually telling his 200MM+ followers how great a deeply broken product is.
It is not possible to make a good version of Grokipedia while satisfying the requirements of its owner, namely to launder ultra-right viewpoints[1].
1: https://www.nbcnews.com/tech/elon-musk/elon-musk-grokipedia-...
Could it be, could it possibly be, that he’s not an honest broker but a deeply wounded, emotionally immature malignant narcissist?
That would require careful planning, introspection, and humility.
What would a "good" version be ?
Start putting real facts into a site and before you know it you're "woke" again with such untruths as Slavery was Bad, Biden won the 2020 Election and of course Full Self Driving is Impossible without Lidar
Seriously. My Christmas wish is for everyone to understand that Grokipedia was not created in good faith, by someone who wants free and open knowledge to flourish.