Abhay Sheshadri

on the atlas — 33

• Inverse Rubric Optimization: A testbed for agent science | Fulcrum2 savers
• Exclusive | White House Reins In AI-Testing Unit as National-Security Concerns Grow - WSJ2 savers
• Aaron Gokaslan on X: "@eliebakouch @PyTorch And this is going to make AI less energy efficient, burn more CO2, and use more water as a result." / Twitter1 savers
• Eliciting bad contexts — LessWrong1 savers
• Book Review: The Dialectical Imagination1 savers
• Postmodernism For Rationalists (my attempt) | Slate Star Codex1 savers
• Automated Weak-to-Strong Researcher15 savers
• Introspection Adapters: Training LLMs to Report Their Learned Behaviors2 savers
• (Some) Natural Emergent Misalignment from Reward Hacking in Non-Production RL — LessWrong1 savers
• Thomas Larsen's Shortform — LessWrong2 savers
• Operationalizing FDT — LessWrong2 savers
• Nyaya1 savers
• AI safety undervalues founders — LessWrong6 savers
• LLMs Don't Need More Parameters. They Need Loops. - YouTube1 savers
• Geoffrey Hinton on X: "@ESYudkowsky LeCun p(doom) = 0.001; Yudkowsky p(doom) = .999; Ensemble p(doom) = 0.5;" / Twitter1 savers
• Is Prosaic Alignment Research Useful If the First Takeover-Capable AI Is Not an LLM? — RedWrong1 savers
• Eliezer Yudkowsky on X: "every man knows in his heart that on the Day of Judgment he shall be redeemed or damned according to how many restaurant ketchup packets he has saved up" / Twitter1 savers
• Recent LLMs can do 2-hop and 3-hop latent (no-CoT) reasoning on natural facts — LessWrong1 savers
• An audio version of my blog post, Thoughts on AI progress (Dec 2025)1 savers
• Eliezer's Unteachable Methods of Sanity — LessWrong9 savers
• Defeating Nondeterminism in LLM Inference - Thinking Machines Lab37 savers
• ALIGNMENT - by vincent huang - a slice of my mind22 savers
• Galaxy brain resistance18 savers
• Guardian Angels: LLM Personalization for Productivity and Security · Gwern.net11 savers
• The stable marriage problem - by Ajeya Cotra - Good Bones10 savers
• 2025 year in review | Kevin Liu10 savers
• Statement on the US government directive to suspend access to Fable 5 and Mythos 5 \ Anthropic9 savers
• Inside vLLM: Anatomy of a High-Throughput LLM Inference System - Aleksa Gordić7 savers
• No-One Escapes the Permanent Underclass7 savers
• llm assistant personas seem increasingly incoherent (some subjective observations) — LessWrong7 savers
• Barnum effect6 savers
• Gwern visits BAIR – Yuxi on the Wired4 savers
• In The Balance | Slate Star Codex2 savers