Aiden Jeong
19 followers · 40 following · 1456 views
on the atlas — 41
- thalassophilia - by hannah - a lot about nothing1 savers
- I don't care (anymore) if your writing got a Pangram false positive7 savers
- i think about it all the time - by Claire Wang11 savers
- The AI Superforecasters Are Here - by Scott Alexander5 savers
- flâneur — a map of Curius links18 savers
- unit-distance-remarks.pdf3 savers
- The Bottleneck8 savers
- The Emperor's Soul.pdf1 savers
- The Central Paradox of Love5 savers
- Before the game begins - by Espen Slettnes - SPARC JC Blog5 savers
- Calvin and Hobbes at Martijn's - Bill Watterson30 savers
- Competence as Tragedy10 savers
- why time felt slower when we were kids (and how to get it back)9 savers
- the art of staying in touch - by maja - velvet noise3 savers
- Keegan: Why we care about whales - Yale Daily News1 savers
- i think about it all the time - by Claire Wang9 savers
- Always invite Anna7 savers
- nothing feels quite like rejection - by vincent huang1 savers
- It's rude to show AI output to people | Alex Martsinovich7 savers
- Mathematical Beauty, Truth and Proof in the Age of AI | Quanta Magazine2 savers
- “Interesting" and "sparkly" - by Andrew Wu - SPARC JC Blog1 savers
- 35 Thoughts About AGI and 1 About GPT-5 - by Steve Newman1 savers
- Looking for Alice - by Henrik Karlsson - Escaping Flatland92 savers
- the key to love is understanding - by maja - velvet noise14 savers
- He Announced His Intention to Die. The Dinner Invitations Rolled In. - The New York Times3 savers
- Opinion | The Paris Métro Smell That Transports Me - The New York Times1 savers
- Explaining the Free Energy Principle to my past self2 savers
- The AI-Box Experiment: – Eliezer S. Yudkowsky3 savers
- Heavy and Light - by Vivian Loh - circles2 savers
- Recursive Self-Improvement - LessWrong1 savers
- AI 20279 savers
- Dario Amodei — On DeepSeek and Export Controls2 savers
- Don’t Surround Yourself With Smarter People54 savers
- Dario Amodei — Machines of Loving Grace34 savers
- Reflections on OpenAI33 savers
- Tracing the thoughts of a large language model \ Anthropic18 savers
- Dario Amodei — The Urgency of Interpretability17 savers
- Six Selections on Self-Defense - by Andrew Wu11 savers
- Mapping the Mind of a Large Language Model \ Anthropic6 savers
- Auditing language models for hidden objectives \ Anthropic4 savers
- CSET-AI-Triad-Report.pdf3 savers
highlights — 61
My solution so far has been to write down the entire mental state you’re in when you truly believe something and go back to it at moments of big decisions.
i think about it all the time - by Claire WangThere’s a sort of person that, upon being faced with a difficult problem, won’t question if they’re the right one for the task but rather believes that, given time and effort, they can figure it out. In direct contrast, there are those who ruminate endlessly, wonder if they’re the right person, if they’re choosing to put time into the right thing… Perhaps, in the most fucking cliche conclusion possible, you kind of have to just do it.
i think about it all the time - by Claire WangShe said I keep relying on changes in my environment to fix my internal issues.
i think about it all the time - by Claire WangYet, it is the only thing you can optimize fo
i think about it all the time - by Claire WangI’m scared of the future version of myself.
i think about it all the time - by Claire Wangsuddenly I was someone to date, someone to talk about feelings with, not someone to go deep on the work with.
i think about it all the time - by Claire WangIn working with AI, for instance, Litt has come to realize that a lot more math than he’d been aware of is about knowing a huge collection of facts and putting them together in interesting ways — rather than having a eureka moment. “I somehow understood that a lot of what I think about as difficult is more about your working memory and knowledge, rather than creativity,” he said. “It shook my confidence that I understand what it is we do or what makes a good mathematician. Now I think that most of what I do is putting together some facts that are known in standard ways. And then every once in …
Mathematical Beauty, Truth and Proof in the Age of AI | Quanta MagazineYou will not be able to explain how this fusion works. So don’t do it.
Looking for Alice - by Henrik Karlsson - Escaping FlatlandIf you want to prompt someone to be authentic and playful and generative, you usually just need to ask them something where they have a rich experience to pull from but have never pulled an answer from that experience before. If you ask two or three increasingly detailed questions about something they tell you, you get there.
Looking for Alice - by Henrik Karlsson - Escaping Flatlandthe conversation is wildly generative for both of you, in that it brings you out, helps you become.
Looking for Alice - by Henrik Karlsson - Escaping FlatlandYou want to rapidly extract as much information as possible, so you can figure out what you like and so that you can pattern match
Looking for Alice - by Henrik Karlsson - Escaping FlatlandI’ve never met anyone who takes his thoughts so with such loving seriosity, and he’s apparently not at all ashamed of his pain. I wish I felt like that about my pain. Also, I need to talk about ideas like this.
Looking for Alice - by Henrik Karlsson - Escaping FlatlandThat is perhaps the most solid dating advice I have, by the way—show the inside of your head in public, so people can see if they would like to live in there.
Looking for Alice - by Henrik Karlsson - Escaping FlatlandAnd then you start hanging out with people like that.
Looking for Alice - by Henrik Karlsson - Escaping Flatlandthat was a mean girl, update the neural net to avoid traits A, B, and C, they go, oh, you know . . . girls. Or, they do the same to themselves: I have a weakness for that type. They also think about relationships in categorical terms.
Looking for Alice - by Henrik Karlsson - Escaping FlatlandAs if I was an untrained neural network trying to predict which image was a cat.
Looking for Alice - by Henrik Karlsson - Escaping FlatlandThe most compelling are the non sequiturs - the mind-expanding ones that make you pause and think wait, what
the key to love is understanding - by maja - velvet noiseOnce you’ve experienced it, you can’t go back to the mimicry, because it feels like a sorry excuse for the real thing.
the key to love is understanding - by maja - velvet noiseAre we risking a little piece of ourselves?
the key to love is understanding - by maja - velvet noise“If afterwards I know nothing more about you than I knew before, then that will be small talk.”
the key to love is understanding - by maja - velvet noiseDid I give them the signal that it’s safe to stop performing and just be?
the key to love is understanding - by maja - velvet noiseIf dialogue is the root of connection, then maybe we should take it seriously, and get better at having them, and at noticing when we’re not. We complain that people are “boring,” without ever wondering if we asked anything interesting. But I think the best conversation is a co-authored rhythm, like a salsa dance, whereas some people treat it like stand-up.
the key to love is understanding - by maja - velvet noiseIf life goes well for me, I hope my dream relationship will just feel like one long, continuous conversation. I don’t mean that metaphorically, but literally — a never-ending dialogue of becoming. I want to walk and talk, lie in bed and talk, sit across from each other at dinner and talk. I want to reflect on our days together, ask strange questions, spiral into metaphysics, laugh about dumb things, and learn the specific, rippled texture of each other’s minds, studying their soul’s cartography, one contour at a time.
the key to love is understanding - by maja - velvet noiseor just played whack-a-mole.
AI 2027For example, research taste has proven difficult to train due to longer feedback loops and less data availability.54
AI 2027because AI systems can eventually help make even smarter AI systems, a temporary lead could be parlayed into a durable advantage
Dario Amodei — On DeepSeek and Export ControlsAll of this is to say that DeepSeek-V3 is not a unique breakthrough or something that fundamentally changes the economics of LLM’s; it’s an expected point on an ongoing cost reduction curve. What’s different this time is that the company that was first to demonstrate the expected cost reductions was Chinese.
Dario Amodei — On DeepSeek and Export Controlsthe shift in the curve due to algorithmic progress is ~1.68x/year
Dario Amodei — On DeepSeek and Export ControlsWhat this typically does is shift the curve: if the innovation is a 2x "compute multiplier" (
Dario Amodei — On DeepSeek and Export Controlsinterpretability commercially to create a unique advantage, especially in industries where the ability to provide an explanation for decisions is at a premium
Dario Amodei — The Urgency of InterpretabilityWe are thus in a race between interpretability and model intelligence.
Dario Amodei — The Urgency of InterpretabilityI consider it basically unacceptable for humanity to be totally ignorant of how they work.
Dario Amodei — The Urgency of Interpretabilityof particular relevance here, some of them productively applied interpretability tools during the investigation
Dario Amodei — The Urgency of Interpretabilityautomate the finding of circuits
Dario Amodei — The Urgency of InterpretabilityWith circuits, we can “trace” the model’s thinking.
Dario Amodei — The Urgency of Interpretabilitycombinations of neurons that did correspond to cleaner, more human-understandable concepts.
Dario Amodei — The Urgency of Interpretabilityoptimized in the slightest to be legible to humans.
Dario Amodei — The Urgency of InterpretabilityThe model uses superposition because this allows it to express more concepts than it has neurons, enabling it to learn more.
Dario Amodei — The Urgency of InterpretabilityThe truth is that science fiction is one of our only sources of expansive thought experiments about the future
Dario Amodei — Machines of Loving Grace) to humans based on some secondary economy of what the AI systems think makes sense to reward in humans
Dario Amodei — Machines of Loving GraceMost people are not the best in the world at anything, and it doesn’t seem to bother them particularly much.
Dario Amodei — Machines of Loving Grace“but with AI’s doing everything, how will humans have meaning?
Dario Amodei — Machines of Loving Gracese. Some early ideas in this direction have been undertaken by the computational democracy project, including collaborations with Anthropic. A m
Dario Amodei — Machines of Loving Gracer example Srđa Popović, who helped to
Dario Amodei — Machines of Loving GraceUnfortunately, I see no strong reason to believe AI will preferentially or structurally advance democracy and peace, in the same way that I think it will structurally advance human health and alleviate poverty. Human conflict is adversarial and AI can in principle help both the “good guys” and the “bad guys”. If anything, some structural factors seem worrying: AI seems likely to enable much better propaganda and surveillance, both major tools in the autocrat’s toolkit. It’s therefore up to us as individual actors to tilt things in the right direction: if we want AI to favor democracy and indiv…
Dario Amodei — Machines of Loving GraceMe: “No! That’s still not radical enough! In the future, that kind of ‘development and integration’ R&D work just won’t need to be done at all—not by humans, not by AIs, not by anyone! Consider that there are 8 billion copies of basically one human brain design, and if a copy wants to do industrial design, it can just figure it out. By the same token, there can be basically one future AGI design, and if a copy wants to do industrial design, it can just figure it out!”
Foom & Doom 1: “Brain in a box in a basement” — LessWrongThese people do not stop existing when camp is over. You can just email them or Discord DM them and schedule a one-on-one (or gasp! meet up in real life) if you miss them.
Opinionated guide to enjoying ratcamps - by Ariel ChengJournal every day. Write stuff down so you don’t forget what it felt like. Also note what you want to do the next day.
Opinionated guide to enjoying ratcamps - by Ariel ChengAnd try to steer past things that you or your conversational partner normally think/talk about. Have conversations that actually prompt novel thought/reflection on both ends.
Opinionated guide to enjoying ratcamps - by Ariel ChengI felt inferior around MOPpers and dismissed my classmates. To other memes I was more resistant: I loved music (and later, writing) too much to be convinced they were not valuable.
Six Selections on Self-Defense - by Andrew Wu