Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Tim Bray on Grokipedia

tbray.org

Post reply on HN
Print view View original

131–140 of 243 posts

Re: Tim Bray on Grokipedia

#131

Post by pstuart »

meowface wrote: post #99
bawolff wrote: post #30

Earlier quoted context omitted.

I think there is a problem sometimes that "debunkers" are often more interested in scoring points with secondary audiences (i.e. people who already agree with them) than actually convincing the people who believe the misinformation. Most people who believe bullshit were convinced by something. It might not have been fully rational but there is usually a kernel of something there that triggered that belief. They also…

This is very, very true. The best debunkers avoid being hostile and make the other side feel like they're being heard and that their feelings and fears are being validated. And they do it in a way that feels honest and not condescending and patronizing (like talking to a child). They make frequent (sincere) concessions and hedges and find as much common ground as they can. Although he's more populist-left and I'm mor…

I'm a fan of Andrew and am impressed by how he's evolved from documenting stupid kids to actually reporting on issues of interest.

I agree that one catches more flies with honey rather than vinegar, but many times it doesn't matter what you say or how you say it -- they're gonna stick to their guns. A prime example of this is in Jordan Klepper interviews where he asks Trump supporters how they feel about something horrible that Biden did, to which they express their indignation; then he reveals that it was actually Trump and they dismiss it because it "doesn't matter".

Re: Tim Bray on Grokipedia

#132

Post by smitty1e »

epistasis wrote: post #116
bebb wrote: post #60

Earlier quoted context omitted.

Because it's a genuinely good idea, and hopefully one for which the execution will be improved upon over time. In theory, using LLMs to summarize knowledge could produce a less biased and more comprehensive output than human-written encyclopedias. Whether Grokipedia will meet that challenge remains to be seen. But even if it doesn't, there's opportunity for other prospective encyclopedia generators to do so.

I don't why an LLM would be better in theory. The Wikipedia process is created to manage bias. LLMs are created to repeat the input data, and will therefore be quite biased towards the training data. Humans looking through sources, applying knowledge of print articles and real world experiences to sift through the data, that seems far more valuable.

> The Wikipedia process is created to manage bias. LLMs are created to repeat the input data, and will therefore be quite biased towards the training data.

The perception of bias in Wikipedia remains, and if LLMs can detect and correct for bias, then Grokipedia seems at least a theoretical win.

I'm happy with at least a set of links for further research on a topic of interest.

Re: Tim Bray on Grokipedia

#133

Post by lschueller »

Grokipedia is a joke. Lot of articles I've checked are AI slop at its worst and at the bottom it says "The content is adapted from Wikipedia, licensed under Creative Commons Attribution-ShareAlike 4.0 License."

Re: Tim Bray on Grokipedia

#134

Post by AgentME »

meowface wrote: post #71
LastTrain wrote: post #51

Earlier quoted context omitted.

“ Grok has historically actually been one of the biggest debunkers of right-wing misinformation and conspiracy theories on Twitter” Well, no, it hasn’t. It has debunked some things. It has made some incorrect shit up. But it isn’t historically one of the “biggest debunkers” of anything. Do we only speak hyperbole now?

I am not using hyperbole or speculating. I absolutely mean it. "Biggest" is tough to quantify, but "most significant" and "most effective" is what I meant. I use Twitter way too many hours a day basically every day and have a morbid fixation on diving deep into right and far-right rabbit holes there. (Like, on thousands of occasions.) Grok is without a doubt the single most important contributor to convincing believe…

> Grok is without a doubt the single most important contributor to convincing believers of right-wing conspiracy theories that maybe the theories aren't as sound as they thought. I have seen this play out hundreds of times. Grok often serves as a kind of referee or tiebreaker in threads between right-wing conspiracy theorists and debunkers, and it typically sides overwhelmingly with the debunkers. (Or at least used to.) And it does it in a way that validates the conspiracy theorist's feelings, so it's less likely to trigger a psychological immune system response.

I've seen this too and agree. It's surprising how well it accomplishes that referee role today, though I wonder how much of that is just because many right-wingers truly expect Grok to be similarly right-wing to them as Elon appears to intend it to be. It's going to be sad when Elon eventually gets more successful at beating it into better following his ideology.

Re: Tim Bray on Grokipedia

#135

Post by morkalork »

beloch wrote: post #129
generationP wrote: post #25

Wondering if the project will get better from the pushback or will just be folded like one of Elon's many ADHD experiments. In a sense, encyclopedias should be easy for LLMs: they are meant to survey and summarize well-documented material rather than contain novel insights; they are often imprecise and muddled already (look at https://en.wikipedia.org/wiki/Binary_tree and see how many conventions coexist without an e…

Bray brought up a really good point. The Grokipedia entry on him was several times the length of his Wikipedia entry, not just because Grok's writing style is verbose, but also because it went into exhaustive detail on insignificant parts of his life simply because the sources were online. My own brief browsings of Grokipedia have left me with the same impression. The current iteration of Grokipedia, besides being un…

>reality control for billionaires

Not to beat a dead horse, but one really could wake up one day and find out we've always been at war with Oceana after the flip of a switch in an LLM encyclopedia.

Re: Tim Bray on Grokipedia

#136

Post by jerf »

jameslk wrote: post #123
f33d5173 wrote: post #118

Earlier quoted context omitted.

It really isn't a promising idea at all. Llms arem't "there" yet with respect to this sort of thing. Having an editor is totally infeasible, at that point you might as well have the humans write the articles.

> Llms arem't "there" yet with respect to this sort of thing Yes, nothing about this is “there yet” which was my point > Having an editor is totally infeasible, at that point you might as well have the humans write the articles. Why?

For the same reason you don't modify autogenerated files in your source code base. It's easy to get an LLM to just regen the page but once someone tries to edit it you're even farther down the road of what an LLM can't do right now. I wouldn't even trust it to follow one edit instruction, at scale, at that size of document, and if we're going to have humans trying to make multiple edits while the LLM is folding in its own improvements... yeah, the LLMs aren't even remotely ready for that at this point.

Re: Tim Bray on Grokipedia

#138

Post by jgalt212 »

> Woke/Anti-Woke · The whole point, one gathers, is to provide an antidote to Wikipedia’s alleged woke bias

According to the Manhattan Institute as cited by the Economist, even grok has a leftwards bias (roughly even to all the other big models).

https://www.economist.com/international/2025/08/28/donald-tr...

Re: Tim Bray on Grokipedia

#139

Post by jameslk »

jerf wrote: post #136
jameslk wrote: post #123

Earlier quoted context omitted.

> Llms arem't "there" yet with respect to this sort of thing Yes, nothing about this is “there yet” which was my point > Having an editor is totally infeasible, at that point you might as well have the humans write the articles. Why?

For the same reason you don't modify autogenerated files in your source code base. It's easy to get an LLM to just regen the page but once someone tries to edit it you're even farther down the road of what an LLM can't do right now. I wouldn't even trust it to follow one edit instruction, at scale, at that size of document, and if we're going to have humans trying to make multiple edits while the LLM is folding in it…

That’s a good point. I think it’s a similar problem of why you wouldn’t let a model go wild in your codebase though. If good solutions to how we handle AI models making code changes are found, it seems reasonable to expect they also may be applicable here

Re: Tim Bray on Grokipedia

#140

Post by siliconc0w »

sunaookami wrote: post #74
siliconc0w wrote: post #61

Not sure it still does this but for awhile if you asked Grok a question about a sensitive topic and expanded the thinking, it said it was searching Elon's twitter history for its ground truth perspective. So instead of a Truth-maximizing AI, it's an Elon-maximizing AI.

This was unintended as observed by Simon here: https://simonwillison.net/2025/Jul/11/grok-musk/ and confirmed by xAI themselves here: https://x.com/xai/status/1945039609840185489 >Another was that if you ask it “What do you think?” the model reasons that as an AI it doesn’t have an opinion but knowing it was Grok 4 by xAI searches to see what xAI or Elon Musk might have said on a topic to align itself with the compan…

The problem is it's part of a pattern of several 'bugs' and even 'unauthorized prompt changes' that have caused Grok to be more Elon-aligned.

And when asked by right wing people about an embarrassing Grok response that refutes their view, Elon has agreed it's a problem and said he is "working on it".

Post reply on HN