Elizabeth Qiu
94 followers · 112 following · 4789 views
on the atlas · 937
- Visualizing Concurrency in Go · divan's blog2 savers
- Takton | Fast & Affordable Manufacturing Tech1 savers
- SASH1 savers
- Engineering Quantum Systems Group - MIT1 savers
- Quantum Coherent Electronics Group at MIT1 savers
- Research ~ CodeDay1 savers
- The components of agent memory \ Anthropic2 savers
- BayMachines1 savers
- Fierce Biotech Homepage | Fierce Biotech2 savers
- 26 Quietly Powerful People Shaping the Future of AI3 savers
- Atomarine1 savers
- Your Claude Code usage2 savers
- What failure looks like - LessWrong15 savers
- How Block Blast harvests its players for ad money · ayoships.com2 savers
- Tacit Labs2 savers
- Hone: AI Engines That Own Outcomes1 savers
- György Ligeti: Lux Aeterna - YouTube1 savers
- Clocktower Bench: AI models playing Blood on the Clocktower2 savers
- How to Secure the DNA Supply Chain | IFP2 savers
- Chanceries | The Diplomatic Atlas1 savers
- Auctions in San Francisco, CA, USA | Maxsold1 savers
- Independent Assessment of AI R&D Acceleration at Anthropic in October 2026 - METR13 savers
- How the world's largest companies acknowledge AI risks | MIT AI Risk Initiative1 savers
- CS283: Governing Artificial Intelligence: Law, Policy, and Institutions1 savers
- The Heilmeier Catechism4 savers
- David_Crane_(politics)1 savers
- Tessellations – The Beneventi School1 savers
- Progress on provable inference for large language models \ Anthropic5 savers
- Logic Analyzers | Saleae2 savers
- Shweta Kukreja on X: "Canva is laying off 22% of its workforce which is 1271 employees, and this is what their CEO said. We need more founders like thissss. https://t.co/YFczt81dyR" / X1 savers
- Stacks Project in Lean1 savers
- 7 Practical Animation Tips4 savers
- ∀ R&D1 savers
- Warp Sciences | AI and robotics for faster drug discovery1 savers
- takeoff-wag-factsheet.pdf2 savers
- How Well Will You Age? Take Our Quiz to Find Out. - The New York Times2 savers
- Digital hygiene | karpathy18 savers
- Home – Lawrence Berkeley National Laboratory1 savers
- Bridge Neurotech: a non-invasive interface for the mind1 savers
- NSF leadership in a new golden age of science | NSF - U.S. National Science Foundation1 savers
- AXION — Non-invasive Neural Interfaces2 savers
- The Council on Strategic Risks - Anticipating, Analyzing, and Addressing Systemic Risks1 savers
- Dilum Sanjaya | Senior Software Engineer & Creative Coder1 savers
- A Narrow Path2 savers
- Goldman Sachs Is Pulling Claude Code From Its Bankers and Traders | State of AI1 savers
- Sparks of Weight-Space Programming: LLMs can natively read & write neural network weights18 savers
- heypcb — AI PCB Designer | Describe it. Get a board.1 savers
- Protocol Explorer · Learn how the internet works1 savers
- GPU-accelerated Weight Teleportation | humans&3 savers
- How Greg Brockman’s Wife Anna Became a Trump and MAGA Mega-Donor | Vanity Fair1 savers
- I don't care (anymore) if your writing got a Pangram false positive9 savers
- Layer Height - OrcaSlicer Wiki1 savers
- Weight-Space Programming from scratch on simple tasks3 savers
- Iraq: The Great Scud Hunt3 savers
- Building Slack2 savers
- Stanford CS 312 | Deep Learning Alchemy8 savers
- PCBs without the magic smoke, using agents8 savers
- sapienne | a community of women building in AI2 savers
- Against SURA — LessWrong13 savers
- Astro Mechanica3 savers
- Home Page1 savers
- bizfileonline.sos.ca.gov/search/business1 savers
- The Metascience Observatory | Analyzing Scientific Reproducibility1 savers
- DC AI Guide — Your Guide to DC's AI Policy Community1 savers
- DC Industrial Policy Salon1 savers
- How To Design A Chip, From Scratch - Jacob Peake4 savers
- Meditations On Moloch | Slate Star Codex36 savers
- Internal Combustion Engine7 savers
- Better MoE model inference with warp decode · Cursor7 savers
- The Sane Asylum - Colossus21 savers
- Researchers at cybersecurity start-up Cathedral got inside Anthropic - The Washington Post1 savers
- Early progress on accelerating drug development with Claude \ Anthropic4 savers
- Pretraining at scale without a pretraining team — Mirendil4 savers
- Resolution: Scale and Automation for Higher Confidence in Alignment — Resolution5 savers
- Laser Cutting and Sheet Metal Services | Fabworks2 savers
- Verification1 savers
- The Ultimate Neolab List3 savers
- What Should the Rules for A.I. Companies Actually Be? - The New York Times1 savers
- What Should the Rules for A.I. Companies Actually Be? - The New York Times2 savers
- RSI vs. Human Enhancement — vie.wiki2 savers
- Reward Misspecification Degrades CoT Monitorability2 savers
- Executive Summary MSR 20251 savers
- How to Run Turing-Complete Malware in PyTorch's Backward Pass | Elements of a Vector Space7 savers
- Empires of AI: The Rise and Fall of a New Global Order3 savers
- ImportInfo | Comprehensive United States Imports Database1 savers
- Mersenne_Twister1 savers
- JLCPCB Factory Live Stream - PCB Manufacturing Process Live Stream | Production Live Stream | JLCPCB1 savers
- Max von Hippel2 savers
- Do LLMs Have Values? A Quantitative Analysis and Alignment Framework for Values in Large Language Models | alphaXiv2 savers
- Antropi Robotics1 savers
- Home | XTX Markets1 savers
- National Compute3 savers
- The Foundation for Science and AI Research (SAIR)2 savers
- You could have invented transformers from memory | Alex Wang13 savers
- Home | Andsoforth1 savers
- Just war theory4 savers
- PostTrain Arena · The open arena for post-training2 savers
- VLR Group1 savers
- Zayd's Blog – Why is machine learning 'hard'?3 savers
- Neuronpedia5 savers
highlights · 258
DAVID CHEN is the son of Argerich's eldest daughter Lida Chen-Argerich. His father is Vladimir Sverdlov-Ashkenazy, the nephew of Vladimir Ashkenazy
pianomania: A PIANO CONCERTO EVENING NOT TO MISS: LEGACY - THE NEXT PIANO GENERATION on 17 March 2026When I Left Big Law, I Learned This
Opinion | What Five Years in Big Law at WilmerHale Taught Me - The New York Timeshe slow layers have the power.
Maintenance: Of Everything, Part OneThe central claim is that Congress’s dysfunction is not primarily a failure of expertise or goodwill, but a failure of internal organization. Factions are one of the few tools available—within existing rules and incentives—to address that failure. The sections that follow are intended to help readers evaluate when factions are likely to be effective, how they interact with parties, committees, and outside groups, and what kinds of support strengthen Congress rather than substitute for it.
Rebuilding Congress from Within: How Factions Facilitate Deliberation and Lawmaking | The Foundation for American Innovationaltruistic people to be more ambitious
Top Performers are Pathologically Ambitious - by Matt BeardTwo scenarios for global AI leadership
2028: Two scenarios for global AI leadership \ AnthropicThe URL shortener that makes your links look as suspicious as possible
CreepyLinkThis emphasis on creating optionality can backfire in surprising ways. Instead of enabling young people to take on risks and make choices, acquiring options becomes habitual. You can never create enough option value—and the longer you spend acquiring options, the harder it is to stop.
The Trouble with Optionality | Opinion | The Harvard CrimsonMaking first contact with superintelligence
Ineffable IntelligenceFinding a friend with only distance and motion.
Marco Polo | Jack HoganJustify anything through the lens of great power competition.
Because ChinaUseful market accuracy improved until early 2025, but hasn’t improved since.
Are Prediction Markets Good for Anything?—Asteriskwhat outcome the costly mechanism would make on the decision if it was called
From prediction markets to info financeif everyone trusts the charts, then anyone with money can manipulate the charts and no one will dare to bet against them. On the other hand, trusting the news entirely is also a bad idea
From prediction markets to info financea betting site for the participants, a news site for everyone else
From prediction markets to info financefirms only benefit from R&D if they are willing to innovate
The Institute Behind Taiwan’s Chip Dominance—AsteriskStudents whose registration at the University of Maryland, College Park, has lapsed for more than 10 years shall be required to complete a minimum of 15 credit hours at College Park after their return to campus in order to earn a baccalaureate degree. This policy can be found in the Undergraduate Catalog.
Re-enrollment Overview | Student Success OfficeThe question left is: how do we close the gap between junior and senior talent?
How do we close the Junior <> Senior talent gap™? · Ady MehtaThere’s also a silver lining to working on disliked products, which is that people only care because they’re using them
Working on products people hateEmail* Please make sure there are no typos Linkedin / website* Github Institution* Primary occupation* Student Academic Founder Industry (part time / full time employment) Other Country of Residence* City of Residence* Tell us about yourself. What’s your most ambitious goal and how does the Encode fellowship help you get there?* What technical project/achievement are you most proud of and why?* Do you have any experience working on cross-disciplinary teams?*
Apply - Cohort 2and gained thousands.
Language Birth—AsteriskI suppose another explanation could be God. I guess this explanation encompasses the first. Through God, I was made lucky. That seems reasonable as well.
Untitledwe do have 45 lb weights.
Post | LinkedInIn our testing, we found that recognizable semantic reconstructions begin to emerge with as little as 15 minutes of data from a new user.
Toward accessible, real-time brain decoding: Introducing ENIGMA | Alljoined BlogI spent maybe in total a month of deep focus on this paper on and off for the course of 7 months, which makes it deeply embarrassing how much I failed to understand probability.
the worst seven months of my life - agniv sarkarDesignating Anthropic as a supply chain risk would be an unprecedented action—one historically reserved for US adversaries, never before publicly applied to an American company.
Statement on the comments from Secretary of War Pete Hegseth \ AnthropicAIs are not capable of directing themselves; most people aren’t either.
Child’s Play, by Sam KrissAI can’t function without instructions from humans, but an increasing number of humans seem incapable of functioning without AI.
Child’s Play, by Sam Krisshe common thread: specificity, accountability, results.
MacKenzie Scott's $26 Billion Sugar Pile | Garry's ListMoney without stewardship is pouring sugar on the floor. You are going to get ants.
MacKenzie Scott's $26 Billion Sugar Pile | Garry's ListCompare that to Warren Buffett, who leads the lifetime giving list at $68.3 billion. He built decades-long relationships with the Gates Foundation and his family charities. He understands what they do. He watches outcomes. Scott’s approach is the opposite — spray and pray at a scale nobody has ever attempted.
MacKenzie Scott's $26 Billion Sugar Pile | Garry's ListA person’s experiences make up 100% of their perception of the world, despite representing only 0.000000001% of what has actually occurred.
EA and AI Safety's Humility Problem — Nixon HannaCaloric and the Utility of Incorrect Theories
Circuits Updates - April 2024Centralization in the political sphere may be inevitable, but certainly not in the technological sphere.
Internetová kontrarevoluceU.S. and allied policymakers tried for years to convince China to play by the WTO’s rules. Not only did those efforts fail, but the United States and its allies now also face a situation in which even if China abided by the rules, other economies would need to violate them to compete. Washington simply cannot diversify its critical mineral supply chain without deploying significant subsidies, tariffs, and other industrial policy tools.
The Case for Upending World Trade | Foreign Affairsneither of the factors that led to the rise of the rules-based trade order exists anymore
The Case for Upending World Trade | Foreign AffairsReagan struck a deal with Japan in which Tokyo committed to “voluntarily” limit its auto exports to the United States in return for avoiding higher tariffs on its cars.
The Case for Upending World Trade | Foreign AffairsWhen unrealistic deadlines become the norm, the team doesn’t actually move faster.
Six Things I Learned Watching a Robotics Startup Die from the Inside | Rui XuOne of the most dangerous ideas going around: robot hardware will become “commodity,” assembled from off-the-shelf parts by Chinese manufacturers just like phones, with the real value sitting in the AI layer.
Six Things I Learned Watching a Robotics Startup Die from the Inside | Rui XuManufacturing is not a service you buy. It’s a capability you build.
Six Things I Learned Watching a Robotics Startup Die from the Inside | Rui XuAnalogies are compression algorithms. They make complex things simple by throwing away information. In a pitch deck, fine. In an engineering decision, the thrown-away information is usually the part that kills you.
Six Things I Learned Watching a Robotics Startup Die from the Inside | Rui XuThere’s this belief going around that AI models are getting so good that hardware can afford to be dumb. Sensors? The model will figure it out from vision. Safety limits? The policy will learn to avoid them. I call this Large Model Chauvinism.
Six Things I Learned Watching a Robotics Startup Die from the Inside | Rui Xu“The whole idea, the reason why our system works well, is that we give scientists money and let them loose. For you to ask us to run this like we would run a weapons program makes absolutely no sense.” Congress would not have it. They said it was totally unacceptable for the agency to run things this way. That was the beginning of the extreme bureaucratization that we have, with hundreds, if not thousands, of laws and regulations, and the insistence that scientists must propose a project and must do the project they said they were going to do.
What’s Wrong with NIH Grants? - by Santi Ruiz - StatecraftClaude is a space to think.
Golden EncountersLet’s take a bad martial artist (let’s call him Agniv Sarkar).
AdviceAddendum: the work is beautiful. I think by the end we can get images of the brain that look like this:
A heart of fireIn past decades, engineers built hardware first and layered software on top. That model is obsolete. Today’s energy systems are co-designed: sensors, power electronics and mechanical structures are specified to enable algorithms that can track a product’s health, prevent failures and optimize output. Hardware–software fusion is the new operating model.
It’s Time to Treat China’s Connected Energy Systems As a National Security Risk | RANDEngineers are impressed by different kinds of work to managers and product managers. I’m most impressed by an engineer who runs down a very difficult bug, or some kind of weird operational issue that was hard to debug. In my experience, managers and product managers are impressed by speed: being able to fix an issue rapidly, or to deliver some piece of UI ahead of schedule.
Trying to impress people you don't respectSo I’m hopeful that the lovable nerds there will be able to present their own smiling optimism to the rest of the world.
2025 letter | Dan WangThe country that is ahead at any given moment will commit mistakes driven by overconfidence, while the country that is behind will feel the crack of the whip to reform.
2025 letter | Dan Wang