Cameron Tuckerman-Lee

Thread on VJEPA 2.1🤟 This DEFINITELY flew under the radar: just a few days ago, @AIatMeta released V-JEPA 2.1, taking a massive step toward closing the gap between image and video domains. For a long time, image backbones were the only option for solving dense vision tasks. This model disagrees, showing that universal spatial understanding also emerges from large-scale video models!🎥

@ziv_ravid Did you read the blog post? They assigned reviewers to 2 groups (based on reviewer preference even!) and only those in group A could not. The detection they did was not sophisticated and based on prompt injections in the manuscripts and humans verified the results.

I (still) wasn't affected by the ICML review policy, which desk rejected all the papers of reviewers who used LLMs to write their reviews (and didn't explicitly mention it) 😱, but this is a bad decision and not a good way to handle AI reviews. First, AI detectors are not reliable enough, with many false positives. Second, if it's a good review, why should I care that AI wrote it? We're using AI assistants everywhere in our day-to-day lives. What is the next step? To ban AI coding agents? I understand the motivation to prevent low-quality reviews, but this is not the way to improve them

👏👏👏

"I'm gonna make him an offer he can't r̶e̶f̶u̶s̶e̶ accept." –V̶i̶t̶o̶ ̶C̶o̶r̶l̶e̶o̶n̶e̶ @ChatGPTapp

@SebastienBubeck @maxwellfarrens I agree with the first sentence of the quote, as do most people I assume, but surely that is orthogonal to whether or not you can force a company to go along with that decision and I don't think the quoted tweet engages with this point in good faith.

@maxwellfarrens x.com/jachiam0/statu…

I sincerely hope this leads to a mass talent exodus from OAI to Anthropic. Until now, it was possible to brush off Anthropic’s principles as empty promises — they sounded good, but in some sense they were still untested. Sure, Anthropic had delayed certain features (like giving Claude access to the internet) while they pursued additional safety testing, but it wasn’t clear if that approach would hold when real pressure arrived. But now the answer is clear: Anthropic is an organization with principles, one that is willing to put their money where their mouth is. Dario will stand up to an administration that has cowed every other notable tech leader into submission, including Sam. I am personally terrified of what AI will mean for the world. I’m not sure anyone, no matter how principled, can effectively steer the ship. But I do know that Sam and OAI definitively cannot. Sam serves Moloch knowingly — he does so with no reservations. So, if I have to entrust this ship to anyone, it’d be Dario. He, along with all the incredible folks I’ve gotten to know at Anthropic, are the reason I still have hope for the future.

Can we stop putting cookie warnings on our web sites or not yet?

@Scott_Wiener @WIRED "The difficulty with this particular rumor is that, while I was unable to substantiate darker allegations, parts of the story still resonate." I heard a rumor that Wired actually practiced journalism but I wasn't able to substantiate that rumor either 🙄

.@WIRED published a tabloid conspiracy piece about how the gays are running Silicon Valley, control everything, use saunas together & discriminate against non-gay people. Buried deep in the article: From 2000 to 2022, a whopping .5% of venture funding went to LGBTQ founders. 🤯

10 years ago, there were 10 people around my kitchen table. No structure, no brand, no certainty. Just a handful of builders showing up for something that yet didn’t exist. Since then @southpkcommons has grown to over a 1000 members. Today, 25,000 people apply to SPC every year. We have offices in SF, NYC, and Bangalore. But SPC isn’t about the numbers, it’s about the people and what emerges when they choose to show up for each other. It looks like @anuraggoel, already successful and credible, showing up with generosity, helping others, sharing work, iterating in public. Patterns sharpened and @render was born. It looks like @thejamescad and @dbabbs, who met at SPC to become cofounders building @tryprofound. One of the superpowers of SPC is the collision rate—not the shallow kind, but the kind that only happens when people keep showing up long enough to build trust. It looks like @MaximilianMona, who moved out to California in an RV to be at SPC and eventually build Ironsite. That’s someone saying, with their whole life: this matters. And it looks like @AshtonJEaton, an Olympic gold medalist, walking into SPC not for a career pivot, but for a deeper reinvention. To trade mastery for learning. Not for optics. For truth. SPC has been designed by the community and for the community, with one goal that hasn’t changed: pay it forward. We’ve helped normalize taking time to find truly meaningful work, whatever shape that might take. So on this ten-year anniversary, I want to thank the people who made SPC what it is. The ones who showed up when they didn’t have a narrative. The ones who lived in the question — and lingered in uncertainty for long enough to find out. The ones who came back, again and again, for the work and for each other. Happy ten years, SPC. Thank you for showing up!

Ryder needs us!

Just updated the sound profile on my @FrameworkPuter 13" and it's like I have new computer. The sound is _so_ much better now it's shocking.

@PalmerLuckey @FrameworkPuter What are you gonna do with all the leftover cash?

Anduril no longer needs to raise another round of funding. I will pay for Arsenal 2 by reselling the the RAM that came in my @FrameworkPuter Desktop a few months ago. Sorry, venture capitalists.

Have you ever done a dense grid search over neural network hyperparameters? Like a *really dense* grid search? It looks like this (!!). Blueish colors correspond to hyperparameters for which training converges, redish colors to hyperparameters for which training diverges.

I can bribe my two year old with way fewer stickers than I expected. The exchange rate is surprisingly favorable.

On full body MRIs: I’ve read this article before and it is not what I would consider “good” Most of the studies he cited were very poorly done and can not be extrapolated at all Most of the arguments are around logistics and potential fear with some random stats on complications Everyone is missing the point. I’ve gotten over a hundred of these scans on patients. Not ONCE have we subjected them to something unnecessary based on the findings. It was a delicate and detailed discussion about risks, benefits, alternatives, how we would cautiously wait and watch in a structured way. On the other side, I’ve caught at least 4 early stage cancers in young people who would’ve never been screened (caught at stage 1 and dealt with), diagnosed years long MSK issues, see early white matter changes that lead to altered cognitive plan, etc etc. Most importantly, I’ve been able to give people piece of mind that nothing huge is underlying (with the caveat we don’t see everything). If it’s about resources and logistics, that’s one thing. If it’s about causing widespread HARM, i just don’t see the data supports that. Why wouldn’t we make the same argument about ALL screening like colonoscopies as well? All of these arguments come down to TIME and TRUST more than anything. If you give your patient time to explain everything, and have a relationship built on trust, then you don’t go spiraling down these pathways, you are informed and empowered by data to make real actionable change in their lives. Also, just to say it, I should be able to get whatever the fuck test I want. The medical establishment has this patriarchal almost priest-like quality of believing only WE can decipher the truth for the plebs. It’s exhausting. And lastly - technology will never stop advancing, it never has and never will. In the next decade, full body MRIs (or some other comprehensive diagnostic) will become so good and so affordable (say $100 each) that it will just be a yearly thing we all do. The science and technology WILL get there, as it always does. But this is the necessary rubicon we must pass. We have to go through this awkward messy phase before we break down paradigms and create a new standard of care. Ok, end rant. And no this post was not sponsored by @prenuvo But you’re welcome @andrewlacy 😀

@MamazMike @Waymo Yeah, Google is really the king of impossibly slow rollouts that are so slow it destroys any excitement they had at launch (see Google+)...

@tuckerman @Waymo Even that is still not showing up on my app, so I stand corrected on what I said

We're officially authorized to drive fully autonomously across more of the Golden State. Next stop: welcoming riders in San Diego in mid-2026! ☀️

@MamazMike @Waymo Last week's announcement was covered under approval they had previously received (I think back in May but I'm not sure). Today's announcement expanded what they have approval for but not their active service area.

@tuckerman @Waymo That came out more than a week ago and it was not covering all they claim in this new post

@MamazMike @Waymo It's not 100% rolled out but some users have down to SJC

@Waymo Then why? Are you guys just announcing for the sake of it?

This week's Hard Fork might be the craziest interview we've ever done. We asked the CEO of Roblox about the child safety issues that have been plaguing the platform. His response was...well, just listen. nytimes.com/2025/11/21/pod…

STARGATE RETURNS! Yes, it’s true.  14 years after the franchise aired its last episode (SGU’s “Gauntlet”), a new Stargate series has been greenlit by Amazon.  And it’s not a reboot or a wholesale reimagining that will wipe the slate clean on 17 seasons and some 350 hours of Stargate history.  It’s a new series that will be the perfect jumping-on point for first-time viewers while, at the same time, honoring the existing past.  And the reason for that is because this new series was created by longtime franchise veteran Martin Gero who worked on SG-1, Atlantis and Universe, writing such notable fan favorites as The Storm, The Eye, Duet, First Strike, Be All My Sins Remember’d and many more. Joining the production of the new series is Brad Wright, co-creator and co-showrunner of all that came before – Stargate SG-1, Atlantis, and Universe.  Martin has kindly offered yours truly a role on the project as well and I have happily accepted. Martin has been developing this show for a while now.  A little over a year ago, he reached out and asked “Want to read something cool?”.  That something, it turned out, was his pilot script for the new Stargate series.  Now I obviously can’t say too much about the content at this point – but I can assure you that it embraces everything that made the original Stargates so great: heart, humor, rich mythology, exploration, action, adventure, compelling/endearing characters, and that overall sense of optimism and fun that made you fall in love with Stargate. You are all in for a treat. Eventually. Now putting this production together is going to take some time even though Martin has already done a lot of the preliminary heavy lifting, writing the pilot script and series overview that details the show, its world and characters, as well as his plans for the first season and beyond.  Between now and the series premiere somewhere down the line, the writers room must be assembled, stories need to be spun, scripts written, prep started, crew hired, actors cast and, eventually, a new Stargate series produced. It will take a while but Martin shares my philosophy when it comes to dialing in the fandom, so expect to receive updates throughout the prep, production, and post process.  Concept art, behind-the-scenes photos, videos and insights, breaking news and much much more! Get ready.  We’re heading through the gate one more time. Chevron Eight is locked!

🤌