Kaylee George
53 followers · 34 following · 2517 views
on the atlas — 137
- A New Era of Midjourney11 savers
- Asymmetry of verification and verifier’s law — Jason Wei15 savers
- Summer Afternoon1 savers
- justinzwu.com8 savers
- cannoneyed.com/projects/isometric-nyc6 savers
- 2025 letter | Dan Wang62 savers
- All the Work That Keeps Us Warm - by Kendall Waldman2 savers
- The Management Secret That Makes SNL’s Chaotic Writers Room Succeed - Fast Company1 savers
- Production Week | Saturday Night Live Wiki | Fandom1 savers
- In pursuit of a benchmark for human taste.2 savers
- AI trading in real markets9 savers
- Solving Regex Crosswords with Z3 - Made of Bugs1 savers
- The 28 AI tools I wish existed10 savers
- On-Policy Distillation - Thinking Machines Lab29 savers
- More Than DNS: The 14 hour AWS us-east-1 outage8 savers
- The Only Important Technology Is The Internet — Kevin Lu17 savers
- 192 Weeks8 savers
- How to Keep Winning11 savers
- Chef Edward Lee's New Korean Restaurant Is Unlike Anything Else in America — Mise En Place - YouTube2 savers
- Inside Victor Wembanyama's summer of transformation3 savers
- A small number of samples can poison LLMs of any size \ Anthropic6 savers
- VaultGemma: The world's most capable differentially private LLM1 savers
- The New World - Colossus24 savers
- AI Market Clarity - by Elad Gil - Elad Blog5 savers
- Import AI 431: Technological Optimism and Appropriate Fear3 savers
- How mega batteries are unlocking an energy revolution2 savers
- From Single-Node to Multi-GPU Clusters: How Discord Made Distributed Compute Easy for ML Engineers1 savers
- All Possible Views About Humanity's Future Are Wild6 savers
- Explaining Gojo's Infinity with Calculus - YouTube1 savers
- Why I’m Leaving Tech to Pursue Art35 savers
- I built ChatGPT with Minecraft redstone! - YouTube3 savers
- Writing Toy Software Is A Joy7 savers
- Bourdain & The Trouble With Being Alive2 savers
- Pete Koomen12 savers
- Why I Lugged My 27-Pound Toddler to a Rocket Launch — Wait But Why2 savers
- Startup Ideas · Gwern.net3 savers
- Humanness in the Age of AI3 savers
- Control as Liability2 savers
- Curius / Onboarding2621 savers
- Staring into the abyss as a core life skill85 savers
- Theory of Change (Aaron Swartz's Raw Thought)83 savers
- The Bitter Lesson78 savers
- Thirty Observations at Thirty62 savers
- i wish we’d grown up on the same advice - by vincent huang61 savers
- How to Pick a Career (That Actually Fits You) — Wait But Why61 savers
- Nat Friedman57 savers
- Don’t Surround Yourself With Smarter People54 savers
- sparkly people and how to find them - by Anson Yu53 savers
- Conviction is the Scarcest Resource: Essays About Gaining Real Conviction45 savers
- 95%-ile isn't that good44 savers
- Why You Should Start a Blog Right Now39 savers
- College as an incubator of Girardian terror | Dan Wang35 savers
- reflections on palantir - Nabeel S. Qureshi35 savers
- The Gap Map34 savers
- The housing theory of everything - Works in Progress34 savers
- What I Miss About Working at Stripe - Every33 savers
- Reflections on OpenAI33 savers
- Impact, agency, and taste | benkuhn.net30 savers
- Conviction is the Scarcest Resource: Dialogues About Conviction30 savers
- World-building IRL - by Joss - box.29 savers
- choosing - by vincent huang - a slice of my mind29 savers
- Works in Progress28 savers
- Dynamicland27 savers
- I am rich and have no idea what to do with my life25 savers
- Questions · Patrick Collison25 savers
- School Is Not Enough25 savers
- Whole Earth Index24 savers
- cdixon | Climbing the wrong hill24 savers
- Notes on “Taste” | Are.na Editorial24 savers
- Whitepaper — The Collective Intelligence Project24 savers
- ambition – A Slice of My Mind22 savers
- No One is Really Working21 savers
- Willingness to look stupid20 savers
- How to Think for Yourself20 savers
- Life is a Picture, But You Live in a Pixel — Wait But Why19 savers
- Tyler Cowen is the best curator of talent in the world19 savers
- By default, capital will matter more than ever after AGI — LessWrong19 savers
- Why and how to write things on the Internet | benkuhn.net19 savers
- On becoming competitive when joining a new company | Ludwig18 savers
- The end of my childhood17 savers
- Startup Advice - Sam Altman17 savers
- Transformer Circuits Thread16 savers
- Putting Ideas into Words15 savers
- ui.land15 savers
- Lessons from Peter Thiel | Posts | 8VC15 savers
- Why I think AI take-off is relatively slow - Marginal REVOLUTION14 savers
- Advice For New College Grads14 savers
- The Top Idea in Your Mind14 savers
- Introducing Zed – A collaborative code editor written in Rust14 savers
- the key to love is understanding - by maja - velvet noise14 savers
- 27 Pieces of Advice on my 27th Birthday - Jackson Dahl13 savers
- Bop Spotter13 savers
- The Moat of Low Status12 savers
- Same Energy | Visual Search Engine12 savers
- Why I Write | The Orwell Foundation12 savers
- PlasticList11 savers
- Century-Scale Storage11 savers
- Failing to Understand the Exponential, Again10 savers
- placestoread.xyz10 savers
- Is My Toddler a Stochastic Parrot? | The New Yorker10 savers
highlights — 131
Summer is so fucking easy; cut a tomato, put some salt on it — you’re a genius, life is wonderful.
All the Work That Keeps Us Warm - by Kendall WaldmanFirst, all the members of the good teams spoke in roughly the same proportion, a phenomenon the researchers referred to as “equality in distribution of conversational turn-taking.”
The Management Secret That Makes SNL’s Chaotic Writers Room Succeed - Fast Company“SNL only works when we have different writing and performing styles all bumping into and meshing with each other,” he said. “That’s my job: To protect peoples’ distinct voices, but also to get them to work together. I want to preserve whatever made each person special before they came to the show, but also help everyone be sensitive enough to make the rough edges fit. That’s the only way we can do a new show every week without everyone wanting to kill each other as soon as we’re done.”
The Management Secret That Makes SNL’s Chaotic Writers Room Succeed - Fast Companythe models still require a smooth curriculum in order to learn. Pretraining is more forgiving because its objective is dense; but to make up for this, RL must use a dense curriculum.
The Only Important Technology Is The Internet — Kevin LuWhat mattered was packing our data into the models as efficiently as possible, and not only did these methods achieve better results, but they seemed to improve with scale
The Only Important Technology Is The Internet — Kevin Luputting something back in the pool of human experience
How to Keep WinningHow could you ever win if you're not in the mindset of winning. If you're not locked in?
How to Keep WinningHe wanted to go, he said, to put his mind and body under unusual stress, to increase his range of motion, to add strength, flexibility and balance through different exercises.
Inside Victor Wembanyama's summer of transformationHe is so intentional in the variety of ways that he tries to improve as a player and person," Johnson said. "It's constantly trying to push himself out of his comfort zone and learn new things. Sometimes, the levels he goes [to] and the things he thinks about that may be a small nuance, but he feels like if I can add this to my world being, mindset, [or] whatever it may be, it can help."
Inside Victor Wembanyama's summer of transformationhe's never witnessed such perspective, or curiosity, from such a young player.
Inside Victor Wembanyama's summer of transformationThe only thing that ultimately matters is how you feel about yourself. Trying to appease others is never a winning game
The New World - Colossus“My advice to you and your team is to just be yourselves,” Rubin explained. “Because if you do what you think the rest of the world wants, one of two things will happen. You might be right, but if you are, you won’t really know why; and if you’re wrong, you’ll be angry at yourself for deviating from what is true to your core. So just be yourself, and either you’ll be right, and it’ll feel really good; or you’ll be wrong. But at least you’ll still be yourself.”
The New World - ColossusLLMs can be poisoned to exfiltrate sensitive data when an attacker includes an arbitrary trigger phrase like in the prompt. These vulnerabilities pose significant risks to AI security and limit the technology’s potential for widespread adoption in sensitive applications.
A small number of samples can poison LLMs of any size \ AnthropicYou can stop a technology if there’s a cheap substitute for it (e.g, nuclear, where other power sources substitute). But it’s very rare to stop a technology that has no substitute, with human cloning perhaps the only exception.
Import AI 431: Technological Optimism and Appropriate FearI should state clearly that I love the world and I love humanity.
Import AI 431: Technological Optimism and Appropriate Fear"I'm just afraid that if I died today, then my life would have amounted to nothing" - Soul, Pixar
Why I’m Leaving Tech to Pursue ArtAt least one model will match the performance of human experts across many industries before the end of 2026.
Failing to Understand the Exponential, AgainWhat to work on. Changing your prioritization is often the single biggest lever for improving your leverage!
Impact, agency, and taste | benkuhn.netwhere is your taste the best?
Impact, agency, and taste | benkuhn.netA common trait of high-agency people is that they take accountability for achieving a goal, not just doing some work.
Impact, agency, and taste | benkuhn.netTaste: you need a good intuition for what things will and won’t work well to try
Impact, agency, and taste | benkuhn.netpeople’s biggest bottleneck eventually becomes their ability to get leverage—i.e., to find and execute work that has a big impact-per-hour multiplier.
Impact, agency, and taste | benkuhn.net“I’m only interested in people as long as they are unpredictable to me. If I can predict what you’ll say, I’ll lose interest. The people I stay close to are the ones who regularly surprise me.”
the key to love is understanding - by maja - velvet noiseThe Greeks, of course, knew this already and had separate words for separate forms: Philia, for deep friendship. Eros, for romantic and sexual passion. Agape, for selfless unconditional love. Storge, for familial affection.
the key to love is understanding - by maja - velvet noiseBut the most intriguing idea to me is the notion of the expanding circle of sympathy: the realization that until recently, most people, even the most enlightened intellectual leaders, simply didn’t see all of humanity as fully human.
The most peaceful time in historyHaber-Bosch process: it turns pure nitrogen and hydrogen gas into ammonia
Turning air into breadSpeed of learning beats perfection every time. Build fast, learn fast, iterate fast. Everything else is just ego.
Velocity is everything · Sid's BlogWhile building Claude Code, I built the todo list feature in 6 hours, dogfooded it internally for a week, then launched externally while continuously making improvements to the UX.
Velocity is everything · Sid's BlogIf I think of ideas where I've sketched out a post in my head but haven't written it down, the number must well into the thousands. If I were to quit my job and then sit down to write full-time until I died, I think I wouldn't run out of ideas even if I stuck to ones I've already had. The world is big and wondrous and fractally interesting.
Some thoughts on writingMy feeling is that the world is so full of interesting stuff that ideas are everywhere.
Some thoughts on writing“decel” is a beautiful turn of phrase.
are you high-agency or an NPC? - by Jasmine SunThe goal is not to avoid mistakes; the goal is to achieve uncorrelated levels of excellence
Nat FriedmanEarly in your career, the most important thing you can do is find excellence as quickly as you can and surround yourself with it.
Life lessons after one year at RostraWe are looking to achieve the goals A, B, C, under the constraints X, Y, Z. No system existed to do so when we set out on our efforts (otherwise we probably would have used it!), so from our experiences building systems in this area, we can distill these experiences into a specific set of system requirements. These requirements may not be obvious to readers who have not spent considerable time thinking about or building applications in this space.
What Makes a (Graphics) Systems Paper BeautifulI think that culturally, we care a bit too much about productivity in the U.S., especially in places I've lived recently (NYC & SF). But at a personal level, higher productivity doing work or chores doesn't have to be converted into more work or chores, it can also be converted into more vacation time or more time doing whatever you value.
95%-ile isn't that goodmost people would blame external factors (lag, luck, a glitch, teammates, unfairness, etc.) when they "died" in the game. The most striking thing about that was that people who did that almost never became good and never became great. I got pretty good at the game3 and my "one weird trick" was to think about what went wrong every time something went wrong and then try to improve.
Willingness to look stupidBefore I publish a new essay, I read it out loud and fix everything that doesn’t sound like conversation. I even fix bits that are phonetically awkward; I don’t know if that’s necessary, but it doesn’t cost much.
Why and how to write things on the Internet | benkuhn.netWriting without examples tends to be dry and abstract.
Why and how to write things on the Internet | benkuhn.netAm I convinced the work is based on a compelling set of goals and constraints? What is the central insight or organizing principle being proposed? What are the benefits the system provides its users? (Do I agree they are valuable?) Can I think of alternative (e.g., simpler) solutions that might be a preferred way to meet the stated goals and constraints? Am I convinced the design decisions made are responsible for successfully achieving the stated goals? Does the system provide the community with a new capability that was not possible (or too difficult) to do before? What are the implications …
What Makes a (Graphics) Systems Paper BeautifulI wonder if it’s possible to be happy with the life we have while aspiring to a better one at the same time
50 Things I Wonder - by Stephen Long - The BackwoodsThe goal is not to avoid mistakes; the goal is to achieve uncorrelated levels of excellence
Nat FriedmanMomentum is critical. Don’t lose it.
Startup Advice - Sam AltmanObsess about the quality of the product.
Startup Advice - Sam AltmanThe women he has kids with are leaders in a host of different fields - Grimes is a famous musician, Ashley St. Clair is a famous influencer, Shivon Zillis is a famous venture capitalist. And Musk has enough money that he can afford to give his hundred kids a $3 billion nest egg each. Start with Musk genes, add some other form of talent, keep remixing them until you get super-embryos, give them insane amounts of starting capital. He’s building a new ruling class for humankind.
Press Any Key For Bay Area House Party - by Scott AlexanderThe statistical and the beautiful cannot coexist
32 notes on AI & writing - by Jasmine Sun - @jasmineWithout wind, the Biosphere 2 trees grew tall and beautiful but fundamentally weak. They had no roots deep enough to support their own growth.
Home | Substackdiscipline matters more than motivation
Home | SubstackTending a garden requires effort, but you get to eat the tomatoes.
Home | SubstackEverything good requires tending. Everything beautiful demands maintenance.
Home | SubstackIf you want to understand a place, you need to walk it
How to build the perfect city