Your curated collection of saved posts and media
Source: https://t.co/2veRRwNUkL
openai projects $115b investment surge by 2029, tripling previous estimates to fuel ai business innovation Source: https://t.co/hR1hqSZUyD
Source: https://t.co/3Q2ganvx9D
IBM only holds a two-to-four percent market share, which might mean they need to cut costs on free services or maybe they're just not focused on attracting users who want that kind of support. Source: https://t.co/Ua2xvUZYSh
🔥 Kai just went LIVE with the Private Sale! From Beverly Hills to the blockchain – trading profits now fueling the future. 💎 👉 Join the exclusive sale here: https://t.co/iXWvTPcav7 #Crypto #PrivateSale #DeFi #Kai https://t.co/zBqIMtFvlL
🔥 Win $KAIS Tokens – Invite Your Friends! 🔥 ✨ Earn extra entries for every referral 💰 Prizes: •50,000 $KAIS tokens – 10 Winners •100,000 $KAIS tokens – 5 Winners 👉 Join here: https://t.co/0gYhutZof5 #Airdrop #Crypto #Giveaway #Solana #Sol #Blockchain #Airdrop #Ai https://t.co/Z4erlZ10GX
🔥 赢取 $KAIS 代币 – 邀请你的朋友！🔥 ✨ 每推荐一位好友可获得额外抽奖机会 💰 奖品: • 50,000 $KAIS 代币 – 10 名获奖者 • 100,000 $KAIS 代币 – 5 名获奖者 👉 立即加入: https://t.co/YZOYfoBHur… #空投 #加密货币 #赠品 #Solana #Sol #区块链 #空投 #Ai https://t.co/erow5bzQlN
🔥 Win $KAIS Tokens – Invite Your Friends! 🔥 ✨ Earn extra entries for every referral 💰 Prizes: • 50,000 $KAIS tokens – 10 Winners • 100,000 $KAIS tokens – 5 Winners 👉 Join here: https://t.co/YZOYfoB9ET… #Airdrop #Crypto #Giveaway #Solana #Sol #Blockchain #Ai @kaisvaulthq https://t.co/Me7FNybAiX
🔥 $KAISトークンを獲得 – 友達を招待しよう！🔥 ✨ 招待ごとに追加エントリーを獲得 💰 賞品: • 50,000 $KAIS トークン – 10名様 • 100,000 $KAIS トークン – 5名様 👉 今すぐ参加: https://t.co/YZOYfoBHur… #Airdrop #Crypto #Giveaway #Solana #Sol #Airdrop #Ai #Coin @kaisvaulthq https://t.co/7JyWGkq64h
There's a lot going on RN on https://t.co/kp5bn0MAjC and making small PRs like this fixing broken links and typos are a HUGE help https://t.co/d0v0bX3Lhi TY to everyone in the community who takes a bit of timeout to push these and improve the exp ❤️
There's a lot going on RN on https://t.co/kp5bn0MAjC and making small PRs like this fixing broken links and typos are a HUGE help https://t.co/d0v0bX3Lhi TY to everyone in the community who takes a bit of timeout to push these and improve the exp ❤️
Platinum Sponsor @solana needs no introduction: you already know it delivers speed, scale, and ultra-low fees across DeFi, NFTs, payments, and consumer apps. Discover what’s coming at TOKEN2049 Singapore, 1-2 October. https://t.co/GoqgZNGOSn
OpenAI is in a full-blown safety crisis. Experts warned from 2017–2022 this exact harm would come. OpenAI ignored the risks and pushed ChatGPT anyway. Now lives are at stake, cases are spiraling, and they are quietly removing features to cover it up. https://t.co/biYsTYvItm https://t.co/SetKrlK32F
Stay calm: a fresh IRL is about to hit. Snag the front row before it disappears ↓ Our handpicked, premium-free in-person jams are straight-up legendary. https://t.co/IYMYBVhc2c Got a rogue project itching to escape? Builders, founders, and creators are logging in. Send it. https://t.co/FtcLDu9XOl
Is #AI Contributing to Rising Unemployment? St. Louis Fed finds "a striking correlation between AI’s prevalence and unemployment increases since 2022." #RiseoftheRobots https://t.co/bx2WkiC9Hn
(4/N) 🔍Feature #1: Large-Scale Demonstration Dataset • 50 tasks, 10,000 demos, a total of ~1,200 hours of data • Subtask and skill (30+) segmentation • Spatial relation annotation • Multi-granularity language annotation https://t.co/XZDl3aptuU
(5/N) 💯Feature #2: High-Quality Data ✔️Teleoperated using our JoyLo interface ✔️Near-optimal, clean demos ✔️Consistent manipulation behaviors ✔️Moderate, consistent teleoperation speed 🚫 No sudden accelerations/decelerations 🚫 No failed grasps 🚫 No unintended collisions https://t.co/kLNUdwYAE2
(6/N) 🏠Feature #3: Long-Horizon Mobile Manipulation in Realistic Homes • Task durations range from 1 to 25 minutes (average 6.6 minutes) • Performed in household-scale scenes • Requires memory, planning, and reasoning over a long period of time https://t.co/LD2ObgdsxT
(7/N) 🧩Feature #4: Diverse State Transitions and Manipulation Skills • Spatial: next_to, inside, on_top, under, touching • Particles: covered, uncovered • Thermal: hot, cooked, on_fire, frozen • Others: open, closed, on, off, attached, sliced, diced https://t.co/AS8KBroBxZ
Can AI help understand how the brain learns to see the world? Our latest study, led by @JRaugel from FAIR at @AIatMeta and @ENS_ULM, is now out! 📄 https://t.co/y2Y3GP3bI5 🧵 A thread: https://t.co/SMevqm5Xxx
Why did we buy India’s most expensive number? Because India deserves its own AI. Sovereign AI is a bigger priority than even nuclear energy in ways we can’t yet imagine. China is already releasing powerful open-source models, but every single one of them calls Arunachal Pradesh “South Tibet.” That’s how subtle propaganda seeps in. Relying on the US or any other foreign country is no safer. That’s why we’re building Puch AI: India’s first AI, the true gateway to AI for a billion+ people. Over the last decade, India has witnessed three defining shifts: Jio made the internet accessible, UPI revolutionized payments, and Quick Commerce redefined shopping. The next leap is mass AI adoption. Why begin with WhatsApp? Because for most Indians, apps are friction. After WhatsApp, we launched AI through voice calls. Will it always be free? Yes, even though many warned us that if it’s free, people would dismiss it as “second-class.” For us, accessibility and empowerment come first. A few days ago, I was in an Uber and asked the driver to try Puch. He thought for a while and then asked: “What’s my wife doing at home right now?” That’s India, curious about AI, but still new and unsure where to begin. Today, nearly a million people use Puch every month, but here’s the part I love most: more than 60% are experiencing AI for the very first time. In small towns, people use Puch in their own languages, often through voice notes, because that’s how they naturally communicate. For first-time users, AI has to feel exciting and magical, not lifeless. That’s why we introduced stickers just like ChatGPT did with Ghibli, brought old photos to life through video, and now done something similar for voice calling: just pick up the phone, call Puch on WhatsApp or SIM, and find out. The obvious crab mentality question: Can Puch AI compete with OpenAI, Meta, and Google? Despite minimal resources, and even with WhatsApp’s restrictions, Puch already offers better video generation than Grok (sorry, Elon) and stronger multilingual support than Meta AI, all of it fully self-hosted without using any external APIs. If we wanted, we’d love to beat them at their game. But the bigger game is having them come to us for distribution. That’s the only game I’m playing, and the one I intend to win. There’s plenty of news about OpenAI, Google, and Meta trying to expand into India. But if we can reject an acquisition worth millions, if one of our own employees can turn down a personal offer from Sam Altman, the only way they beat us here is by bribing government officials. Nothing else would be enough. Real behavioral change, and yes, unlimited iPhone giveaways, requires capital. If you only move when others clap, watch this and cheer from the sidelines: https://t.co/isLra3NFdk. But if you want to win the fight for the world's biggest AI market, let’s talk. PS: If you tag an investor whose money we accept, you win an iPhone.
Nine former CDC directors who served under Republican and Democratic presidents warn that the damage RFK Jr is doing to America's public health system is "unlike anything our country has ever experienced." https://t.co/T5hPhH4VeK
A classic paper, collab between @AIatMeta , @GoogleDeepMind , and @NVIDIAAIDev Language models keep personal facts in a measurable amount of “storage”. This study shows how to count that storage—and when models swap memorization for real learning. 📡 The Question Can we separate a model’s rote recall of training snippets from genuine pattern learning, and measure both in bits? The paper says Yes. Encode the dataset twice: once with the target model, once with a strong reference model that cannot memorize this data. The extra bit-savings the target achieves beyond the reference counts as rote memorization. The shared savings reflect genuine pattern learning. 🧮 The Measurement Trick Treat the model like a compressor. If a data point becomes shorter when the model is present, those saved bits reveal memorization. Subtract the savings that also appear in a strong reference model; the remainder is “unintended” memorization. 📏 What the Numbers Say GPT-style transformers store roughly 3.6 bits per parameter before running out of space. Once full, memorization flattens and test loss starts the familiar “double descent” curve right when dataset information overtakes capacity. 🔄 Why It Matters Loss-based membership inference obeys a clean sigmoid: bigger models or smaller datasets raise attack success; large token-to-parameter ratios push success to random chance. Practitioners can now predict privacy risk from simple size ratios instead of running attacks. Most interesting nugget: a single scaling rule—bits per parameter—explains capacity limits, double descent, and membership-leak risk all in one shot.
🌀Diversity Aware RL (DARLING)🌀 📝: https://t.co/MH0tui34Cb - Jointly optimizes for quality & diversity using a learned partition function - Outperforms standard RL in quality AND diversity metrics, e.g. higher pass@1/p@k - Works for both non-verifiable & verifiable tasks 🧵1/5 https://t.co/AhEYPQwbkg
Starting the semester with @kchonyc telling the story of @NYU_Courant & @NYUDataScience CILVR group to a full house of students. Featuring @rob_fergus & @ylecun. https://t.co/Uj7sm9u0Z5
John Oliver fact checking RFK Jr. https://t.co/0MsrO4lqNC
John Oliver fact checking RFK Jr. https://t.co/0MsrO4lqNC
1/ You might have seen it—DINOv3 is out! 🦖🦕In this thread, we share key insights on our Gram anchoring ⚓︎ and how it helps to get smooth feature maps. 👇 https://t.co/694H7CWdIN
In my latest article, "If We Build It, Everyone Lives," (https://t.co/OEkx9Juu0W) I walk through the many ways AI is helping the world already and why Doomer fears have a a 0% success rate and yet they keep getting press. They move the chains and promise us that this time they'll be right and we'll all be very sorry real soon, just you wait. If their crazy policy suggestions get passed, like the MIRI report (https://t.co/BXRosrNsKp) that calls for mass government surveillance of AI researchers and restricting GPUs like they're AK-47s, they will do real, lasting, nasty damage that's infinitely worse than the thing they're supposed to stop or prevent. That's because when activists push big, scary headlines about the bad things they predict a technology will bring, a silent spring (https://t.co/aifHo06jOr), mass unemployment, a new ice age (https://t.co/CHDTFrWsuu), they usually dead wrong and they almost always ignore the good things we stand to lose without the technology: the jobs that never get created, the clean air we don’t breathe, the cascade of new inventions that never come to be. Their fears mask the miracles already happening every day. AI is already unleashing a torrent of potential breakthroughs from fighting malaria (https://t.co/3yWbk0paDV), to battling Parkinson's, to enzymes that munch plastic to discovering new batteries and much more. We seem to resent progress today, to think that economics is a zero sum game where when we make things more efficient everyone gets poorer when it works in the exact opposite way. It's a belief echoed by AI pioneer and socialist turned doomer, Geoffrey Hinton, who recently told the Financial Times (https://t.co/vfj40Zmd1T) "AI will make a few people much richer and most people poorer." He also said a decade ago that radiologists will be made obsolete by AI but we have widespread adoption of AI by radiologists instead and more radiologists than ever so I wouldn’t give his predictive powers much credibility at all. More than likely, AI isn't a replacement for human intelligence; it's an amplifier, a co-intelligence. People who embrace AI and work closely with it will displace people who refuse to work with it. Film editors who refused to switch to digital editing don't have a job anymore but there are more editing jobs than ever. That's because making films got easier with digital tools and that meant more films, which meant more jobs. Give people power tools and they are remarkably adaptable with them. It's the beginning of a great co-creation, a partnership between human and machine that will unlock unprecedented levels of creativity and productivity. AI is a general purpose technology. It's not a sentient being with its own desires. It’s a tool. It's a mirror that reflects our own ingenuity, our own compassion, our own desire to build a better world. To fear it is to fear ourselves. To demand we stop building is to demand we stop dreaming, stop striving, and stop reaching for a better future. The future is not a predetermined path to destruction. It's a landscape of infinite possibility, and we are the architects. We can choose to be guided by fear, to huddle in the dark and curse the light, or we can choose to build. We can choose to use these incredible new tools to solve the oldest of human problems: disease, poverty, and ignorance. So let the doomers have their bleak fantasies. Let them wallow in their apocalyptic predictions. We have work to do. We have a world to build. And in the world we’re building, AI isn’t the end of humanity. It’s the beginning of a new chapter, one where we are healthier, smarter, and more capable than ever before. If we build it, everyone lives. And we are building it, right now.
No freakin way 🤯 https://t.co/e4qolDpYp2
No freakin way 🤯 https://t.co/e4qolDpYp2
BREAKING: @AnthropicAI has raised a $13B Series F round led by @ICONIQCapital, with @Fidelity and @lightspeedvp as co-leads. This round raises their valuation to $183 billion. https://t.co/KVS3pEcjBf
