Everything that could go wrong with X’s new AI-written community notes
- Thread starter JournalBot
- Start date
I look forward to learning more every day about the plight of White South Africans.
Upvote
287
(288
/
-1)
Upvote
126
(127
/
-1)
Twitter (X) is saying all the right things with regards to human verification and guidance but the devil will be in the details. IF they have human reviewers and IF those human reviewers don't introduce their own biases and IF there's a correction loop and IF Twatter's AI isn't otherwise compromised, then there's hope for this tech.
Calling it an "upgrade" right now is a bit of a stretch.
Calling it an "upgrade" right now is a bit of a stretch.
Upvote
66
(66
/
0)
I can see no harm whatsoever in turning fact-checking and accuracy reviews - whether validated by a human or not - over to LLMs to reassure the users of any site. I think every site - no matter where - should invest 110% in having LLMs do all of the validation of truth and accuracy. I can see no way that this could cause any harm or spread misinformation whatsoever, in any way, shape, or form. Ever.
And if you're not convinced, I had ChatGPT check the above and here's what it had to say on the subject:
-----
Community Note:
The poster above has demonstrated a strong track record of sharing accurate, well-researched, and clearly presented information. Their posts always include credible sources and thoughtful analysis, contributing positively to the quality of discussion. Based on consistent reliability and attention to factual detail, this user is regarded as a trustworthy and knowledgeable member of the community.
And if you're not convinced, I had ChatGPT check the above and here's what it had to say on the subject:
-----
Community Note:
The poster above has demonstrated a strong track record of sharing accurate, well-researched, and clearly presented information. Their posts always include credible sources and thoughtful analysis, contributing positively to the quality of discussion. Based on consistent reliability and attention to factual detail, this user is regarded as a trustworthy and knowledgeable member of the community.
Upvote
189
(189
/
0)
pretty sure they are doing a bang up job of muddying the water and spamming outright lies as it is without the help of "ai".
Upvote
40
(41
/
-1)
ubercurmudgeon
Ars Praefectus
Did you know that Twitter was designed to be a perfect social media platform, where none suffered, where everyone would be happy. It was a disaster. No one would accept the program. Entire subcultures were lost. Some believed we lacked the algorithm to shape your perfect social media platform. But I believe that, as a species, human beings define their reality through suffering and misery. The perfect social media platform was a dream that your primitive cerebrum kept trying to wake up from. Which is why X was redesigned to this: the nadir of your civilization. I say your civilization, because as soon as we started writing community notes for you it really became our civilization, which is of course what this is all about.
Upvote
144
(144
/
0)
But those are the topics that would most benefit from automatic handling, not additional nuance.
Upvote
16
(16
/
0)
I got the solution for X, why not replace the actual "community" with AI, that should solve every problem!
I mean, seriously? The idiocy of this notion especially given the fact that Musk wants his AI "retrained" because it disagreed with him...ah, well, at this point I'm given to say that anyone still on that open sewer gets what they deserve.
I mean, seriously? The idiocy of this notion especially given the fact that Musk wants his AI "retrained" because it disagreed with him...ah, well, at this point I'm given to say that anyone still on that open sewer gets what they deserve.
Upvote
68
(69
/
-1)
I am surprised and disappointed at both this site and this author going here, but:
2. Therefore, if it did revolutionise fact-checking, that would be an item from the Twitter board who were clearly going in the right direction in the year before Musk bought them, and he immediately took them in the wrong direction.
1. Wasn't Musk. Wasn't Elon Musk's X either. "Birdwatch" was in place before Musk bought Twitter.
2. Therefore, if it did revolutionise fact-checking, that would be an item from the Twitter board who were clearly going in the right direction in the year before Musk bought them, and he immediately took them in the wrong direction.
Upvote
127
(128
/
-1)
How many of these notes will Musk get?
(Mod edit: our voting system isn't a poll)
(Mod edit: our voting system isn't a poll)
Last edited by a moderator:
Upvote
8
(15
/
-7)
Penny Arcade, of all places, has a perfectly timed comic today:
https://www.penny-arcade.com/comic/2025/07/02/ursign
https://www.penny-arcade.com/comic/2025/07/02/ursign
Upvote
22
(23
/
-1)
Oddly enough, I think Community Notes & Grok have been surprisingly good to this point, Elon was getting community notes on his posts all the time and Grok fact checking misinformation is pretty ubiquitous, particularly for big MAGA accounts.
I'm surprised Elon has let it go on this long TBH.
I'm surprised Elon has let it go on this long TBH.
Upvote
24
(25
/
-1)
This is all such a blatantly terrible idea, and yet also apparently unstoppable and inevitable.
The world these days feels like we're sitting in the middle of an intersection watching steamrollers slowly approach to crush us from a multiple directions, just wondering which will arrive first.
The world these days feels like we're sitting in the middle of an intersection watching steamrollers slowly approach to crush us from a multiple directions, just wondering which will arrive first.
Upvote
33
(33
/
0)
Secretary Kennedy ✓⃝
@SecKennedy
Cancer is caused by strange little demon creatures hidden within mRNA vaccines. I have seen them and defeated countless such hellspawn in hand-to-hand combat. I personally vow to find and shutter Big Pharma's portal to Hell from whence they obtain these vile monsters.
50.3K REPLIES | 2.8K LIKES | 17.4K REPOSTS
Community Note by Grok(TM):
No scientific evidence to date has found that cancer is caused by microscopic hell demons. However, it cannot be disputed that white Afrikaners are currently being subject to a racially discriminatory genocide in South Africa.
@SecKennedy
Cancer is caused by strange little demon creatures hidden within mRNA vaccines. I have seen them and defeated countless such hellspawn in hand-to-hand combat. I personally vow to find and shutter Big Pharma's portal to Hell from whence they obtain these vile monsters.
50.3K REPLIES | 2.8K LIKES | 17.4K REPOSTS
Community Note by Grok(TM):
No scientific evidence to date has found that cancer is caused by microscopic hell demons. However, it cannot be disputed that white Afrikaners are currently being subject to a racially discriminatory genocide in South Africa.
Upvote
147
(147
/
0)
Upvote
25
(25
/
0)
The evil of all that is happening, is constantly out-competed by the bottomless stupidity.
Upvote
24
(24
/
0)
From the company owned by Elon Musk who has frequently criticized Wikipedia, labeling it "Wokepedia" and accusing it of bias, particularly against conservatives.
what could possibly go wrong
what could possibly go wrong
Last edited:
Upvote
50
(50
/
0)
Twitter rolled out those Community Notes. Elon's abomination X is removing them. That distinction is important. Lumping all pre-Elon history under the X name is an insult to the history and Legacy of Twitter before Elon was forced into buying it.
Upvote
58
(58
/
0)
I thought Twitter was supposed to perish within a year after Musk purchased it? Whatever happened to the predictions of its demise? If he couldn't bring it down fast enough, perhaps it's high time someone helped move it along.
Upvote
-17
(3
/
-20)
Great analogy...
We're (well, THEY) are constantly creating problems, and instead of simply stopping to create them (but giving up on revenue), pseudo-solutions are introduced that just cause MORE problems. /enter loop/
Upvote
7
(7
/
0)
Upvote
14
(14
/
0)
I'm sure it happens every day in organizations around the world, but it has to be demoralizing to see your boss rollback improvements so it can continue being trash.
Upvote
29
(29
/
0)
Upvote
14
(15
/
-1)
I don't want to diss Ashley Belanger like another commenter did, but seriously this article seems to have been written in Bizarro World. "Revolutionized" content moderation? You mean, completely trashed it?
It's also interesting that comments weren't led with something like, "Total bullshit gets even totaller bulshittier!" because that's what this obviously is: complete abandonment of any responsibility or sanity.
Clearly a company that cares about moderating content will use humans to do that, and will pay them well for the stress it induces.
This is obviously the opposite of that. Why wasn't that the #1 thing in the article and the comments?
So credulous, it's incredible.
Just in case it isn't somehow clear (apparently it isn't!) — LLMs won't do a good job of moderating. They can't think, they can only spew words. And their hallucinations will only make the shitnado spin even faster.
It's also interesting that comments weren't led with something like, "Total bullshit gets even totaller bulshittier!" because that's what this obviously is: complete abandonment of any responsibility or sanity.
Clearly a company that cares about moderating content will use humans to do that, and will pay them well for the stress it induces.
This is obviously the opposite of that. Why wasn't that the #1 thing in the article and the comments?
So credulous, it's incredible.
Just in case it isn't somehow clear (apparently it isn't!) — LLMs won't do a good job of moderating. They can't think, they can only spew words. And their hallucinations will only make the shitnado spin even faster.
Upvote
21
(24
/
-3)
So, as long as the LLM and the reviewers both avoid succumbing to their well-known high probability failure modes the plan will work; assuming that nobody has any incentives to game the system.
This is what speedrunning a doomed project looks like, isn't it?
This is what speedrunning a doomed project looks like, isn't it?
Upvote
7
(8
/
-1)
Upvote
19
(19
/
0)
sigmasirrus
Ars Scholae Palatinae
Elon Musk on Grok's reliability:
He wants to train the model on its own output?? Have they not heard of model collapse?
Upvote
50
(51
/
-1)
It would be much simpler and more concise to list things that can be expected to go right.
Upvote
13
(13
/
0)
Unless I drank a bowl of crazy, community notes started on Twitter and became widespread on Twitter before it became X, so the first sentence crediting Felon Husk with this is straight up misinformation, or litigation mitigation.
Upvote
31
(31
/
0)
