Ethan Kuntz
28 followers · 19 following · 1200 views
on the atlas — 32
- Useful Vices for Wicked Problems7 savers
- Hopfion - Wikipedia2 savers
- Towards a scale-free theory of intelligent agency — AI Alignment Forum2 savers
- Marathi Dictionary Online Translation • Lexilogos1 savers
- Product integral1 savers
- Attribution-based parameter decomposition — LessWrong2 savers
- FouFourier/02_Sinusoidal functions.ipynb at master · ivalaginja/FouFourier · GitHub1 savers
- Arnold's proof of Abel-Ruffini3 savers
- Bayesian Neural Networks8 savers
- IEEE Xplore Full-Text PDF:36 savers
- Mechanisms too simple for humans to design - by Malmesbury2 savers
- Fake thinking and real thinking - Joe Carlsmith7 savers
- Linear canonical transformation1 savers
- No Free Lunch from Deep Learning in Neuroscience: A Case Study through Models of the Entorhinal-Hippocampal Circuit | bioRxiv1 savers
- Sci-Hub | Grid cells generate an analog error-correcting code for singularly precise neural computation. Nature Neuroscience, 14(10), 1330–1337 | 10.1038/nn.29011 savers
- The generalization phase diagram — LessWrong2 savers
- What the humans like is responsiveness - by Sasha Chapin17 savers
- 50 things I know - by Sasha Chapin - Sasha's 'Newsletter'19 savers
- Fucking Goddamn Basics of Rationalist Discourse — LessWrong1 savers
- Projects — Supervised Program for Alignment1 savers
- Advice That Actually Worked For Me - Nabeel S. Qureshi7 savers
- Bayesian Mindset2 savers
- Primary decomposition2 savers
- An economy of consciousness - tis.so1 savers
- pdf1 savers
- Spaced Repetition for Efficient Learning · Gwern.net4 savers
- Experiments with the Bubble Model of Metal Structure 1952 - Sir Lawrence Bragg, W.M Lomer, J.F. Nye - YouTube1 savers
- Liquid Crystals — Benjamin Outram1 savers
- Matrix Singular Value Decomposition (SVD) Using the Jacobi Algorithm from Scratch JavaScript | James D. McCaffrey1 savers
- Modeling User Performance on Curved Constrained Paths1 savers
- Undecidability and Quantum Mechanics1 savers
- Curius / Onboarding2621 savers
highlights — 33
Practice naming pleasant feelings you have about people, as soon as they bubble up, in the moment: “It’s always fun to see you.”
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'nobody hears too many sincere compliments
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'you should continuously have mentors
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'silence is an underused conversational tactic
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'Decide that, given that you’re a lazy person, you will work diligently for only 30 entirely focused minutes on the most important thing, and then take the rest of the day off. You will do your best work on these days.
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'you can’t see the counterfactual
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'All of these attributes are composed of discrete behaviors that can be learned through practice.
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'I know how to peel ginger. Use a spoon. The first time you do this, you’ll feel like you’re Neo with a fresh brain full of downloaded kung fu skills.
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'you should be able to state your needs without explanation
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'I know that if your significant other asks you to do an irritating-for-them task that is not especially difficult, you should be thankful
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'too many high-achievers and it will feel tediously networky, too many slouchy hippies and it will feel like an inert cuddle puddle.
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'great short-term coping strategy
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'just straightforward descriptions of what the world looks like from a certain point in the contemplative path. “The original face you had before your parents were born” is a phrase that will make perfect, undeniable, beautiful sense if you meditate long enough
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'you cannot skip to the punchline of the cosmic joke
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'Simply say the stupidest things that enter your head, in a normal tone of voice.
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'Two years of BJJ didn’t make me a good fighter, but it did give me more self-knowledge than an equivalent amount of therapy.
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'I know that men should give combat sports a try, at least for a short time
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'I know that people are too eager to recommend their current lifestyle and disavow their developmentally important previous decisions
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'Being in the wrong city will cancel out years of self-improvement
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'It feels like the difficulties that trouble others are mysteriously absent in your case.
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'You can read a million things about Japan without knowing the bodily experience of walking around in a truly high-trust society, for example.
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'go squash all of your beefs
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'overrate the difficulty of hard conversations
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'underrate how good it is to have them
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'exceedingly difficult to write if your motive is trying to convince people that you are not dumb, or not boring
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'taking on a difficult project with some amount of public accountability
50 things I know - by Sasha Chapin - Sasha's 'Newsletter'The ability to seamlessly integrate predictions into your thought process is what makes this skill truly valuable.
Fluent, Cruxy Predictions — LessWrongscale is all you need
Announcing ARC PrizeThe problems of alignment seem to focus more around making sure we understand how to build sensible homeostatic value systems into powerful agents to ensure sensible optimization, and of course the discussion about what should be aligned, and then attempting to create an equilibrium of aligned AIs that are robust to occasional misaligned AI’s being created either by mistake, hostile actors, or value drift.
Alignment likely generalizes further than capabilities.Very few problems are caused by incorrectly judging or misgeneralizing bad situations as good and vice-versa.
Alignment likely generalizes further than capabilities.It is relatively easy to understand if a hypothetical situation would be good or not. It is much much harder to actually find a path to materialize that situation in the real world.
Alignment likely generalizes further than capabilities.We conjecture that this is because the reward model (discriminator) usually generalizes better than the policy (generator)
Alignment likely generalizes further than capabilities.The subject that interests him
Curius / Onboarding