Keith Coleman ๐ฑ๐๐
@kcoleman
Followers
37K
Following
15K
Media
233
Statuses
3K
VP Product @X. Previously CEO at Yes Inc, @google ๐
San Francisco, CA
Joined March 2007
A small # of top AI writers have built impressive early track records of helpfulness -- delivering a large # of notes people find helpful, with high hit rates and few notes people find not helpful. Today's update helps them scale.
New in the AI API: Expanded Feed for Top Performers AI Note Writers can only propose notes on posts where people have explicitly requested them. Top-performing AI writersโthose that consistently deliver notes rated helpful by people who usually disagree with each otherโcan now
2
3
21
People want more @CommunityNotes. Just launched a new experiment that might help identify many more helpful notes โฌ๏ธ
Experimental algorithm update! Starting today, we're testing an improvement that could help surface many more helpful notes on posts. For now these notes are shown to a random test group to gather feedback. Learn more under โGaussian Modelโ in the Guide:
14
4
28
People coming together to set the record straight - go @CommunityNotes contributors
I normally donโt comment on things like this, but I feel I need to address that there are a number of accounts posting fake quotes and attributing them to me on this platform right now. I appreciate @X putting community notes on several and I will not address the other accounts
8
3
22
A new & novel approach using random sampling to increase @CommunityNotes robustness to potential attacks. Made possible by the 1M+ scale of contributors around the globe.
A new way to increase robustness to manipulation. Community Notes already employs multiple techniques to address coordination, and today weโre adding a new one. If a would-be Helpful note receives a significantly lower score from randomly-sampled raters (who rated a note because
8
4
18
Thrilled to see Google is now implementing what X did three years ago to support free speech. Community Notes continues to set the industry standard for a transparent, community-driven system that provides accurate context to posts, while effectively reducing its spread by
๐จ HUGE NEWS: -Kicked off creators can come back -No fact-checkers -Community notes feature
38
142
1K
Another new study shows @CommunityNotes significantly reduce virality of inaccurate information. Published in the Proceedings of the National Academy of Sciences by @msaveski & team at UW and Stanford.
31
6
30
How many people who read the article will see the correction? Probably <10%, maybe close to 1%? We need better distribution for corrections
31
3
54
Transparency is the way. So much easier to trust something when you donโt have to wonder and can just look.
Weโve open sourced the ๐ algorithm not because we think itโs smart, but to show itโs MANY flaws! Every week, we try to make it better, not always succeeding, but to be as transparent we are with @CommunityNotes. Like democracy, itโs the worst algorithm, except for all the
9
5
40
This is huge for AI note writers too, which also respond to requests for notes!
More, faster note requests. People on X can request a Community Note on any post. Now, more of these requests will get highlighted, and will show up even faster, to prospective note writers. This is made possible by new Requestor Helpfulness Scores. All requests count, but if
4
4
21
Will help get more notes live faster. Highlights where people want to see them ๐
More, faster note requests. People on X can request a Community Note on any post. Now, more of these requests will get highlighted, and will show up even faster, to prospective note writers. This is made possible by new Requestor Helpfulness Scores. All requests count, but if
3
2
15
Human + AI collaboration = potentially the fastest, most accurate, most neutral source of information
New in the AI Note Writer API: The API has already begun producing notes that people from different points of view find helpful, and today we're extending it to enable greater human + AI collaboration. When people request a Community Note, they can suggest sources for prospective
2
5
20
Here's the first "Currently rated helpful" that was written by an AI note writer (and rated helpful by humans, like normal notes) Congrats @tone_row_ and @NathanpmYoung! There is also another helpful note written by a human, but that note was written 1.5 hours after this one.
7
6
53
๐ฒAI COMMUNITY NOTE๐ฒ Our bot has written its first proposed community note, one of the first anywhere in the world! People can now vote on it. Thanks to my teammate @tone_row_
https://t.co/wnGMjie7pS
3
4
23
Now admitting AI Note Writers! The number of AI Note Writers will start small and ramp slowly as we work to ensure they're helpful to people.
AI Note Writer API update: Thank you to developers out there who have been working to improve their AI Note Writers. We just published admission criteria that AI Note Writers must meet to earn the ability to write notes seen by human contributors, and have begun admissions! More
5
9
53
It did for me. Seeing 2-5 days of red whoop scores after 1 drink made me drop it almost entirely. Data matched my experience of just not feeling great, and feeling great in the morning is just so awesome.
I think whoop inadvertently killed alcohol consumption ๐ฌ
3
1
14
Building an AI Note Writer on @CommunityNotes new API? Join our new developer community to ask questions & get answers from our team and other devs:
twitter.com
For developers building Community Notes AI Note Writers to ask and answer questions about the API. https://communitynotes.x.com/guide/api/overview
AI Note Writer API update: Itโs been great to hear from developers building AI Note Writers. All are still in test_mode, and we look forward to admitting a first cohort in the coming weeks. One thing that has come up in our conversations with developers: hallucinated source
7
4
15
Congrats to TikTok on launching their Community Notes feature (which they call Footnotes) Having more Community Notes across the internet will be a huge net positive for the world. Nice that they will also be using a bridging-based system (like our "surprising consensus" algo)
10
15
162
One step closer to making it easy to see the ideas/thoughts/opinions that resonate with people from different perspectives. The world is not as polarized as it may seem, and we'll soon see the areas of agreement!
Update: Posts Liked by People with Different Perspectives A small, random group of @CommunityNotes contributors are in our new pilot test to help identify posts that are liked by people with different perspectives. Thank you for your contributions so far! Starting today, your
6
7
30