Jay Baxter
@jaybaxter
Followers
6K
Following
11K
Media
160
Statuses
2K
@CommunityNotes Founding ML Lead. Member of Technical Staff @X / @xAI. Built BayesDB @MIT
Joined February 2011
The edit experience in Grokipedia v~0.2 is already quite nice, and will keep improving rapidly!! IMO the best part is how it transparently logs its reasoning for why it accepted or rejected each edit suggestion. In the future, with upvotes and downvotes on suggestions, there
https://t.co/op5s4ZiSwh is now version ~0.2 and the early beta version of proposed edits is now live
6
5
35
For many tasks we use grok internally, 4.1 is a massive improvement on our evals. Even bigger than jump from 3->4
We definitely a bit lowkey this time. No ads no hype, but clearly from our user metrics and feedbacks - ppl love it and growth is very strong. 4.1 is a great model.
15
44
502
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 writersāthose that consistently deliver notes rated helpful by people who usually disagree with each otherācan now
2
3
22
This will mean more good notes, faster šš
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 writersāthose that consistently deliver notes rated helpful by people who usually disagree with each otherācan now
1
1
29
@aypan_17 and I went pretty deep on post-training for this one, plenty of room for improvement on the safety front but please try it out and share your feedback with us!
Introducing Grok 4.1, a frontier model that sets a new standard for conversational intelligence, emotional understanding, and real-world helpfulness. Grok 4.1 is available for free on https://t.co/AnXpIEOPEb,
https://t.co/53pltyq3a4 and our mobile apps. https://t.co/Cdmv5CqSrb
2
4
55
There's a lot of highbrow misinformation
NYTimes article quotes someone saying they are "terrified" of Waymo in paragraph 6. Waits until paragraph 33 (out of 44 paragraphs) to mention that they are 91 percent safer than human drivers. How outraged would liberals be if a news outlet covered vaccines like this?
6
21
323
Distraction free reading Available now on the š app for iOS
59
23
606
Should we make an XBridgers handle and quote these or something Whatās a non-shit name for this
@nikitabier @banger Posts on controversial topics that get surprising agreement across both sides of the political spectrum!! https://t.co/6kb41jhHuU
11
0
22
Working on showing more notes!!
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 āGaussian Modelā in the Guide:
6
2
43
What a roller coaster
Update 2: even correcting for the fake views, traffic to Substack links from X is up substantially. (full post reads, signups, etc. also track.) We're so back!
3
0
15
Shucks
Update: most of the apparent lift is fake, unfortunately. Looks like the new browser is pre-loading the links before the user sees them. Iāll update when we know more.
2
0
11
Oh yeah, it cost $0 beyond having a premium+ subscription already btw!
0
0
4
Yes, instance of https://t.co/qkM39iqDVE
New study of 10K twins finds the rise in autism diagnoses is not matched by a rise in autism symptoms. As Iāve long said, autism rates are increasing not because of vaccines or Tylenol, but because of an ever widening definition of autism. https://t.co/kLagSj1NPC
2
5
73
follow up - we had an incredible group of fellows create 20+ projects for improving epistemics and coordination. aiforhumanreasoning dot com
We're launching the FLF Incubator Fellowship on AI for Human Reasoning! š We're seeking talented researchers and builders to develop AI tools that enhance epistemics and coordination. 12 weeks, $25k-$50k stipend. Applications due June 9th.
1
1
14
Hilarious to see this "DEBUNKED" stamp on a *proposed* note that wasn't showing except to contributors. Their own screenshot plainly says the note is "not shown on X" and "needs ratings" Indeed, the whole reason we have a rating system before showing notes is because proposed
Mistake in AI-generated Community Note sparks controversy about anti-Trump protest ā”ļø https://t.co/5PImJtLLr9
11
1
53
I personally strongly dislike posts like this that restate a post but provide zero additional value beyond it. Very bad etiquette to not quote, link, or even screenshot the original: https://t.co/sRkeI8xqtG
JUST IN: Elon Musk says they are working hard to fix the X algorithm. "My apologies for frustrations with the X algorithm. We are working hard to fix the problem."
150
20
885
Researchers take note: this is way way cheaper than before if all you want to do is fetch all the posts with community notes
This is literally how xAI works. You can just do things. I built consumption-based billing for the X API from scratch because I thought it was important enough. No annoying bureaucracy or corporate politics. And you can just do things too if you join us š
9
8
143