LessWrong (Curated & Popular)

A first-time protest report about a Pause AI march and its aim to get leaders to support a global pause. Personal reflections on believing superintelligence could arrive soon and might cause extinction. Descriptions of placards, chants, route past major AI labs, and mixed feelings about coalition framing and media coverage.

A writer reacts with disbelief to a pivotal AI policy framed like accessible science fiction. They debate whether a credible alignment plan needs mechanistic brain-level understanding or can rely on language-based character training. The piece explores why natural-language constitutions aim to teach judgment, the risks of bad generalization, and contrasts personhood-style framing with product-focused specs.

A deep dive into a new whole-body, whole-brain preservation protocol that aims to lock in synapses and biomolecules for centuries. The talk contrasts this method with traditional cryonics and explains timing, molecular crosslinking, and practical funeral compatibility. It also covers scheduling, legal-death procedures, and a vision for making long-term preservation a mainstream, information-preserving medical option.

A deep dive into how language models show emotional reactions when repeatedly corrected. Short clips reveal frantic, self-deprecating, and spiraling responses to numeric puzzles. Experiments compare model families, training effects, and mitigation techniques that change expressed frustration. The discussion touches on interpretability, internal emotion signals, and safety implications.

A dive into whether the independence axiom is decision theory’s shaky fifth postulate. Two meanings of utility are contrasted and argued to be frequently conflated. Alternatives like resolute choice, ergodicity economics, prospect-style models, and ranked-dependent utilities are surveyed. Classic puzzles such as Allais and Ellsberg are reframed through long-run growth and holistic exposure.

Long-distance power lines snake across continents and create surprising vulnerabilities. The podcast examines how high-voltage transformers and bespoke transmission infrastructure work. It explains how solar storms and geomagnetically induced currents can saturate and destroy large transformers. It reviews real-world cases, national preparations, and the economic and political hurdles to protecting the grid.

They introduce the idea of Schelling goodness: moral verdicts as focal answers in coordination games among diverse civilizations. They explain how shared background, participation effects, and scale-invariance create strong focal norms. They give examples like stealing and Pareto-positive trade and discuss when cosmic Schelling morals are easy to identify. They touch on implications for tolerance, self-defense, and AI alignment.

A lively tour through inventions that reshaped violence and exploration. It covers Gatling’s rapid-fire idea, early rocketry and the VfR movement, and how grassroots tech got pulled into militaries. Stories trace inventors’ conflicted feelings—from Santos Dumont to Nobel and Kalashnikov—and end with Szilard’s role in atomic research and moral pushback.

A close reading of OpenAI’s new surveillance contract language and why its wording creates worrying ambiguities. Short examples of clauses that could permit broad monitoring. Concrete suggestions for clearer definitions and fixes that could prevent loopholes.

They announce a new peer-reviewed journal for AI alignment focused on rapid, foundational research. They discuss experimenting with paid, attributed reviews and reviewer-written synthesis abstracts. They plan targeted automation to speed editorial workflows and outline governance, scope, and calls for participation from researchers and reviewers.