LessWrong (30+ Karma)

podcast

Audio narrations of LessWrong posts.

#192 US · Overall 2223 recordings 2025 → today Daily updated 3h ago transcripts through December 2025

All episodes

901–920 of 2223

“How much should the ideal person cry wolf?” by KatjaGrace contributed by Lukas · May 1

“Conditional misalignment: Mitigations can hide EM behind contextual cues” by Jan Dubiński, Owain_Evans contributed by Lukas · May 1

“Risk from fitness-seeking AIs: mechanisms and mitigations” by Alex Mallen contributed by Lukas · May 1

“Sanity-checking “Incompressible Knowledge Probes”” by Sturb, LawrenceC contributed by Lukas · May 1

“AI unemployment and AI extinction are often the same” by KatjaGrace contributed by Lukas · May 1

“AI risk was not invested by AI CEOs to hype their companies” by KatjaGrace contributed by Lukas · May 1

“Cyborg evals” by Eye You, frmsaul contributed by Lukas · Apr 30

“To what extent is Qwen3-32B predicting its persona?” by Arjun Khandelwal, ryan_greenblatt, Alex Mallen contributed by Lukas · Apr 30

“Research Sabotage in ML Codebases” by egan contributed by Lukas · Apr 30

“Maybe I was too harsh on deep learning theory (three days ago)” by LawrenceC contributed by Lukas · Apr 30

“Notes on Transformer Consciousness” by slavachalnev contributed by Lukas · Apr 30

“On today’s panel with Bernie Sanders” by David Scott Krueger contributed by Lukas · Apr 30

“No Strong Orthogonality From Selection Pressure” by lumpenspace contributed by Lukas · Apr 30

“Learning zero, and what SLT gets wrong about it” by Dmitry Vaintrob contributed by Lukas · Apr 30

“The Most Important Charts In The World” by Zvi contributed by Lukas · Apr 30

“LLM Style Slop is Absolutely Everywhere” by silentbob contributed by Lukas · Apr 29

“Goblin Mode, 24 Hours Later” by Dylan Bowman contributed by Lukas · Apr 29

“Let Kids Keep More Productivity Gains” by jefftk contributed by Lukas · Apr 29

“llm assistant personas seem increasingly incoherent (some subjective observations)” by nostalgebraist contributed by Lukas · Apr 29

“Not a Paper: “Frontier Lab CEOs are Capable of In-Context Scheming”” by LawrenceC contributed by Lukas · Apr 29

page 46 of 112

module · podcast

#192 Overall · United States

charted in 25 countriesFull rankings →

module · podcast

226 episodes · last 30 days · avg 4 min

Wed 372Fri 357Thu 338

Episodes per week over the last six months.

module · podcast

Audio narrations of LessWrong posts.