curius graph

all pages

showing 69201-69250 of 202639 pages (sorted by popularity)
Sequoia sempervirens - Wikipedia
Monte Carlo - Wikipedia
Berkeley Marina - Wikipedia
Z-Library - Wikipedia
Wayback Machine - Wikipedia
If Anyone Builds It, Everyone Dies
I enjoyed most of IABIED — LessWrong
AI Optimism – For a Free and Fair Future
Why it's hard to make settings for high-stakes control research
Why imperfect adversarial robustness doesn't doom AI control
Emil Ryd
Golden ratio base - Wikipedia
The Thinking Machines Tinker API is good news for AI control and security — LessWrong
Against Muddling Through — LessWrong
When Is Insurance Worth It? — LessWrong
Covert Malicious Finetuning — LessWrong
Reducing risk from scheming by studying trained-in scheming behavior — LessWrong
Dwarkesh Patel on X: "The most interesting part for me is where @karpathy describes why LLMs aren't able to learn like humans. As you would expect, he comes up with a wonderfully evocative phrase to describe RL: “sucking supervision bits through a straw.” A single end reward gets broadcast across https://t.co/lYonLgrukB" / X
When does training a model change its goals? — LessWrong
IMO challenge bet with Eliezer — LessWrong
AI stocks could crash - Benjamin Todd
Alex Bores, Assemblyman, Joins Primary to Succeed Nadler in Congress - The New York Times
Rationalists Are Less Credulous But Better At Taking Ideas Seriously — LessWrong
How Well Does RL Scale? — LessWrong
The costs of caution
Toy models of AI control for concentrated catastrophe prevention — AI Alignment Forum
What I learnt from having a brain tumor - by Amy Deng
Comments - AI is probably not a bubble - by Peter Wildeford
AI is probably not a bubble - by Peter Wildeford
Critiques of the AI control agenda — LessWrong
trees are harlequins, words are harlequins — welcome to summitbridge, an extremely normal...
evhub's Shortform — LessWrong
Daniel Kahneman: Beware the ‘inside view’ | McKinsey
Sputnik crisis - Wikipedia
Ben's public dating profile
How to survive until AGI — LessWrong
Mourning a life without AI — LessWrong
Hideyness - by Gretta Duleba - Letters to Boys
What training data should developers filter to reduce risk from misaligned AI?
The Problem — LessWrong
Getting 50% (SoTA) on ARC-AGI with GPT-4o
LessWrong
Be Naughty — LessWrong
Alignment By Default — LessWrong
Three positive updates I made about technical grantmaking at Coefficient Giving (fka Open Phil) — LessWrong
Meta-level adversarial evaluation of oversight techniques might allow robust measurement of their adequacy — LessWrong
Nose / throat treatments for respiratory infections • Otherwise
Deciding whether to parent Archives • Otherwise
Some data from LeelaPieceOdds — LessWrong
Drake Thomas