Your curated collection of saved posts and media
https://t.co/ySk3oH9OgM
yeah it’s fucking good https://t.co/8FOQMNLC2p
Nous poursuivons l'installation du wifi très haut débit à bord de nos avions, permettant de profiter gratuitement d’une qualité de service stable, rapide et sécurisée. En cette fin d'année, déjà 30% de notre flotte est équipée et la totalité le sera fin 2026. https://t.co/NfLNy9U1I3
🚨 TESLA + STARLINK = INTERNET EVERYWHERE (YES, EVEN IN THE MIDDLE OF NOWHERE) Tesla just got the green light from the FCC to use Starlink terminals on moving vehicles… aka, satellite internet in your car is officially on the way. Basically, Tesla filed a patent covering a vehicle roof assembly that is transparent to radio frequencies, specifically noting that it allows for satellite communications to pass through. That's important, as signals from SpaceX satellites can't reliably penetrate glass or metal. This project isn’t theoretical either, it means Teslas could soon have an always-on, high-speed internet powered by SpaceX’s global Starlink network. Forget losing signal in remote areas, your car could stream Netflix in the desert, get OTA updates in the mountains, or run Full Self-Driving with real-time data literally anywhere on Earth. If you thought Teslas were already futuristic, this is next level. We still don't know when or if this is coming to Tesla vehicles, but the patent filing is our best sign yet that it's coming soon. Source: InsideEVs, @tesla @Starlink
Once it tips, it will happen fast. There will be attempts to outlaw the use of AI, but the economic incentives will be too powerful. Nations will have to figure out what to do with a wave of unemployed, formerly high-income members of the priestly caste who have no prospects. https://t.co/Eozy4FEuBq
"Grok Heavy is still the smartest AI. I recommend trying it out. Grok 4 Heavy is where we spawn several agents where they work in parallel, and they compare their output like a study group, and give you the final conclusion. And it keeps getting better." 一 Elon Musk https://t.co/toUzDV69DU
GROK TURNS YOUR CAMERA INTO A VISUAL INTELLIGENCE TOOL Finally, Grok made your phone smarter than your last five Google searches. PIC BRAINS: * Snap anything Grok explains it like you’re five * Textbooks, receipts, street signs Grok eats it all * Don’t type just talk Grok gets it instantly * Voice plus vision equals full-blown AI sidekick Time to give your camera something visually intelligent to do. Not only selfies! Source: @xAI, Bloomberg
15u @IndySelect Loading…come out Sunday 12/21 at 2pm at Silver Street https://t.co/GpWBbfvTn3
Efficient GPU code starts with correct data mapping. Our latest GPU Puzzle Tutorial shows how to operate on 2D matrices in Mojo, with a focus on correct thread indexing, bounds checking, and clean memory access patterns. Watch now ⬇️ https://t.co/WaejogfPUA
#Tenpay Checkout hits 92 countries—#WeChat mini programs go global • Merchants accept local currency from Chinese tourists seamlessly • No app downloads—tailored for cross-border shopping • Crushes traditional payment rails in speed/cost • Asia-tech dominates fintech globalization Still have doubts about QR codes? Watch as they go on a world tour! → #fintech #tech #finserv #AI @BetaMoroney @efipm @BrettKing @spirosmargaris @jasuja @enricomolinari @mikeflache https://t.co/T4SLPH5Kmr
My latest blog. Driving Manufacturing Excellence by Unlocking the Power of MOM and MES solutions - @Atos https://t.co/zQaFGwwtby @ArturHabant @elaniazito @IanLJones98 @CurieuxExplorer @Shi4Tech @enilev @jblefevre60 @Fabriziobustama @mvollmer1 @AnthonyRochand @JolaBurnett @lyakovet @debashis_dutta @3itcom @ahier @Analytics_699 @antgrasso @CathCervoni @chidambara09 @DigitalColmer @dinisguarda @DimitriHommel @EvanKirstel @FrRonconi @GlenGilmore @gvalan @HeinzVHoenen @ipfconline1 @jeancayeux @jorgecunha @kalydeoo @nafisalam @Nicochan33 @pierrepinna @PawlowskiMario @puneetsinghal22 @ralph_ohr @RLDI_Lamy @rshevlin @sarbjeetjohal @SpirosMargaris @StefanoDeCupis @tewoz @thomas_dettling @Ym78200 @aure79lien
My latest blog. Driving Manufacturing Excellence by Unlocking the Power of MOM and MES solutions - @Atos https://t.co/zQaFGwwtby @ArturHabant @elaniazito @IanLJones98 @CurieuxExplorer @Shi4Tech @enilev @jblefevre60 @Fabriziobustama @mvollmer1 @AnthonyRochand @JolaBurnett @lyakovet @debashis_dutta @3itcom @ahier @Analytics_699 @antgrasso @CathCervoni @chidambara09 @DigitalColmer @dinisguarda @DimitriHommel @EvanKirstel @FrRonconi @GlenGilmore @gvalan @HeinzVHoenen @ipfconline1 @jeancayeux @jorgecunha @kalydeoo @nafisalam @Nicochan33 @pierrepinna @PawlowskiMario @puneetsinghal22 @ralph_ohr @RLDI_Lamy @rshevlin @sarbjeetjohal @SpirosMargaris @StefanoDeCupis @tewoz @thomas_dettling @Ym78200 @aure79lien
Last week, we had a chance to host @maxhodak_ , founder of @ScienceCorp_ and Co-Founder of Neuralink, for @sv_icons dinner. It was one of the most fascinating conversations we’ve had at Icons, covering topics ranging from the future of brain-computer interfaces to learnings that Max had after co-founding company with Elon Musk. Here are a few insights from the conversation: - Max starts from a blunt premise: the current human condition is mediocre, and the whole point of brain–computer interfaces are not only to treat pathology, but to reengineer the human experience. - He sees the next 10 years as a phase transition: with AI and BCIs compounding at today’s pace, our world will end up either absolutely beautiful or absolutely crazy, but definitely not a slightly different version of 2025. - Max frames AI and BCIs as two distinct technological quests. If the end goal of artificial intelligence is superintelligence, the end goal of the BCI quest is a conscious machine. Classical BCIs are about reading and writing information to and from the brain—extracting motor representations to control a cursor or keyboard, or writing sensory information to restore vision, hearing, or proprioception. But he believes a non-classical form of BCI is on the horizon. Instead of reading or writing information, these systems would enable phenomenal binding over a network. That would be a fundamentally different kind of technology, and if it works, it wouldn’t just improve human capabilities – it would qualitatively transform the human condition. - Reasoning is, in his view, the most important capability humans brains have — and widespread LLM use quietly erodes it, which is why he actively restricts LLMs at work and pushes people to keep doing hard, first-principles thinking themselves. - Max co-founded Neuralink with Elon and speaks very highly of his management skills: when you put extreme urgency at the center of decision making among people with the highest level of critical thinking - impossible things become possible. - One of the big observations that Max had is that the statement of risk is overrated. People can take way higher risks but they don’t have enough agency to think clearly about such risks. He shares the example of Silicon Valley founders - almost everyone can raise money and then if things don’t work - join a big tech firm. No risk is involved. - Max describes a biohybrid approach to BCIs built from stem-cell–derived biological neurons. Instead of placing mechanical or electrical devices into the brain, the idea is to embed engineered neurons into a device and engraft it onto the brain’s surface. The question is whether those cells would grow in and form functional connections – and the answer is yes. Even a tiny graft, on the order of four millimeters by four millimeters, can integrate extensively. Because cells are small, such a device can contain on the order of a million neurons, forming billions of synapses. Max sees this as a genuinely new kind of system, and possibly the first plausible path to a whole-brain – or interhemispheric-bandwidth – neural interface. - Consciousness is independent from intelligence. LLMs have intelligence but don’t have consciousness. Conversely, non-intelligent objects might have consciousness. One of the problems with consciousness is defining it, we all know it exists, but it always has a broad definition. - Causing visual hallucinations via video is possible but the implications are very limited. - During interviews, the most important quality that Max checks is agency. Agency in making all the decisions people made to get where they are or they just followed the flow? - Max’s definition of success for someone is this: does this person have the life they wanted to have? And it doesn’t necessarily have to be rich or luxury life. Someone wants to have good work-life balance and hang out with friends - does this person have this? - His long-term economic thesis is that powerful tech will drive us into a state of abundance, and once that happens, capital and opportunity can be distributed far more equally, making today’s rich-vs-poor discourse feel oddly outdated. - Max thinks about neural engineering from the bottom up: his work is largely in low-level cognition and neural mechanics, even when he talks in terms of high-level concepts like consciousness and identity. - One of the biggest problems to scale BCI is 10 bits-per-second bottleneck of human output. You can increase input, but how can you scale output? It might be possible biohybrid approaches using stem-cell-derived neurons that integrate into the brain, and breaking by directly serializing thought. - He’s deeply interested in the “binding problem” of consciousness — how billions of neurons over space and time become a single unified “moment,” and why you only ever experience your bundle of sensory modes together rather than, say, your vision and someone else’s hearing. - His working hypothesis is that consciousness is tied to physically realized information and thermodynamics: whenever a system burns energy to stabilize a representation against entropy, you get qualia — an idea he’s honest about holding with medium-low epistemic confidence but sees as concrete enough for a real research program. - On ethics and regulation, he argues that concerns around “neural privacy” are often overstated, and that overly restrictive rules around pooling brain data are slowing down progress on exactly the science that could make BCIs safe, useful, and eventually life-changing. Thanks to @CMT_Digital and @silkroadih for supporting the dinner.
This robot solving a rubiks cube in 0.103 seconds is a little preview of what "AGI" really means https://t.co/kskJO2lhT0
You can now adjust specific characteristics in ChatGPT, like warmth, enthusiasm, and emoji use. Now available in your "Personalization" settings. https://t.co/7WSkOQVTKU
🆕 Writing blocks make it easier to craft the perfect email in ChatGPT. ∙Update & format text right in chat ∙Highlight to ask for changes, and accept or reject suggestions ∙Open in your email client once you’re ready to send Try it & please let us know what you think! https://t.co/2Vgf0Av3u6
Open sourcing Tensor2Tensor that has some of the latest sequence modeling techniques! https://t.co/cauSzYBaZB
Slides from my #MTMarathon lecture on "Architectures for Neural Machine Translation"! Link https://t.co/QufCCJuADh and overview https://t.co/fUk4Hxr1MN
7 Types of Artificial #NeuralNetworks for Natural Language Processing https://t.co/ceJUGFo08v https://t.co/UndwyR2AeU
Understanding how #chatbots work is important. Learn the inner workings of an artificial neural network (ANN) for text classification. https://t.co/Z7o6XsxaoA https://t.co/VJc1XvGyRA
Narrative Science Employs Natural Language Generation Image Source: Narrative Science #AI #MachineLearning #DeepLearning #BigData #Fintech #ML #DL #Banking #startups #tech https://t.co/vFaX427aXi https://t.co/HrswFgyQgT
Can we improve chatbots after deployment from natural user feedback? A big YES :) User feedback has rich cues about errors but it cannot be used directly for training. So we use GANs to generate training data from feedback. https://t.co/v6mjQ4J8Ru #Findings-of-#EMNLP2020 #NLProc. https://t.co/j9JHqL8Bgb
Most neurons in language models are "polysemantic" – they respond to multiple unrelated things. For example, one neuron in a small language model activates strongly on academic citations, English dialogue, HTTP requests, Korean text, and others. https://t.co/PrqtDGar0J
We also systematically show that the features we find are more interpretable than the neurons, using both a blinded human evaluator and a large language model (autointerpretability). 📄 https://t.co/XQvzENHMrp https://t.co/dawkxhAvix
Introducing Sora, our text-to-video model. Sora can create videos of up to 60 seconds featuring highly detailed scenes, complex camera motion, and multiple characters with vibrant emotions. https://t.co/YYpOAcrXQ3 Prompt: “Beautiful, snowy Tokyo city is bustling. The camera moves through the bustling city street, following several people enjoying the beautiful snowy weather and shopping at nearby stalls. Gorgeous sakura petals are flying through the wind along with snowflakes.”
😳😳😳 I'm just gonna leave this right here. https://t.co/RvSAsHeXxM
We want to celebrate our heroes on the front lines. Thank you to our frontline workers who worked with the public to get people registered and helped with early voting and polling. Your giving and caring spirit are what Indian Country is all about. #NativesVote https://t.co/GAh3UkdWnA
【好評受注中】 『ぽよよん♥ろっく』さんイラスト 「好久水(すくみず) みどり」 「好久水(すくみず) ぴんく」 https://t.co/ot6bEDoDmr #ネイティブ #エロホビ https://t.co/FadhcrFoLc
【初お披露目】 『左藤空気』さんイラスト 「浅葉 依吹」 企画進行中！ #クレイラドール #エロホビ https://t.co/9uorQyRbsn
gud tek: https://t.co/wOWdglFmPx https://t.co/2mrY6uQrUg
BREAKING: Cursor is acquiring Graphite Cursor CEO @mntruell & Graphite CEO @MerrillLutsky will be live on TBPN at 11:30a PT to break it all down. https://t.co/ASyIHfha9T
I'm leaving @tldraw to enter the world of contracting. From January, I'll be prototyping contributor tools at @wikipedia. My next availability is June! https://t.co/TVbpYLQ1E1