That’s an interesting one, thanks for flagging. None of the things it’s quoting got notes - the C_3C_3 post in the screenshot with the 4.6M views still doesn’t have it. Saad’s post is later than a lot of others, who were writing around when it happened. I wonder if that had something to do with how it was received.
There’s “grok is this true” colloquy (with correct grok responses) in the replies to saad, but that’s also true of other top-level tweets that aren’t Noted. But I do wonder if the circulation of screenshots from “legit” sources just confounds crowd/AI based fact-checking in the early hours. I’ve seen community notes on other politically salient tweets with big partisan lie buy-in (eg covid vax), though ofc you don’t see them uniformly.
Agree, it's worth looking at how the AI-drafted "collaborative notes" evolve if you want to see that in detail. (vs the AI API-submitted notes, where people are responsible for their bots getting it right. Those are 2 different programs). It's definitely harder to reach consensus on the partisan things, which isn't surprising. This one seemed like a layup; the citations in the Saad note are good, they include the US Atty.
https://sciety.org/articles/activity/10.31234/osf.io/85quw_v2 - there are a couple of papers/preprints that compare the Grok-is-this-true volume with CN volume and find Grok seems to be winning out. Which is once again a centralized form of fact-checking, just with a different referee and displayed in the comments rather than on the post.
The notes did appear and grok did update, though. So how long was the lag before this happened? Hours?
The notes did not appear. You can go look at the link I posted. They stayed in No Consensus for days. Last time I checked they were still there.
Grok - once there are reliable sources available for RAG it can get it right.
I checked this morning and saw them on some Top tweets I clicked for query, see eg:
https://x.com/gadsaad/status/2063824627907121430?s=46
That’s an interesting one, thanks for flagging. None of the things it’s quoting got notes - the C_3C_3 post in the screenshot with the 4.6M views still doesn’t have it. Saad’s post is later than a lot of others, who were writing around when it happened. I wonder if that had something to do with how it was received.
That is one of 3 that now shows up as "Rated Helpful" for the keyword "Pratt" https://notetracker.socialmedialab.ca/?start=2026-06-02&end=2026-06-10&filterType=all&q=Pratt The other two are on different topics.
There’s “grok is this true” colloquy (with correct grok responses) in the replies to saad, but that’s also true of other top-level tweets that aren’t Noted. But I do wonder if the circulation of screenshots from “legit” sources just confounds crowd/AI based fact-checking in the early hours. I’ve seen community notes on other politically salient tweets with big partisan lie buy-in (eg covid vax), though ofc you don’t see them uniformly.
Agree, it's worth looking at how the AI-drafted "collaborative notes" evolve if you want to see that in detail. (vs the AI API-submitted notes, where people are responsible for their bots getting it right. Those are 2 different programs). It's definitely harder to reach consensus on the partisan things, which isn't surprising. This one seemed like a layup; the citations in the Saad note are good, they include the US Atty.
https://sciety.org/articles/activity/10.31234/osf.io/85quw_v2 - there are a couple of papers/preprints that compare the Grok-is-this-true volume with CN volume and find Grok seems to be winning out. Which is once again a centralized form of fact-checking, just with a different referee and displayed in the comments rather than on the post.