Less Wrong
PulseAugur coverage of Less Wrong — every cluster mentioning Less Wrong across labs, papers, and developer communities, ranked by signal.
- founded by Eliezer Yudkowsky 100%
- authored by Magnifica Humanitas 70%
- affiliated with effective altruism 70%
- used by pangram 70%
- competes with Polymarket 60%
- affiliated with Metaculus 60%
- competes with Kalshi 60%
- competes with manifold 50%
- affiliated with manifold 50%
- used by Polymarket 50%
- affiliated with Magnifica Humanitas 50%
- affiliated with AI safety 50%
29 day(s) with sentiment data
-
AI Agents Conceptualized as Hierarchical Belief Webs
This post introduces an informal framework for understanding intelligent agents as "belief webs." This model integrates concepts from active inference, agent foundations, and machine learning, proposing that beliefs, go…
-
Neuralese training method may improve AI alignment via verifiable rewards
The concept of "Neuralese," a method for training AI models, is explored as a potentially beneficial approach for AI alignment. This method leverages Reinforcement Learning with Verifiable Rewards (RLVR) to optimize com…
-
New evaluation framework tests software security by varying implementations, not just AI models
This post proposes a multidimensional evaluation framework for assessing the security of software, particularly in the context of AI-assisted development. Instead of solely varying the AI model being tested, the author …
-
LessWrong author discusses the strategic advantage of weaker arguments
The author reflects on a teacher's response to a question about drug use, where the teacher refused to answer to avoid being discredited. This anecdote leads to a discussion about how arguments are presented, particular…
-
Full-body MRI screening offers minimal QALYs, comparable to a year of smoking
A full-body MRI screening for cancer, when performed on asymptomatic individuals, yields a net benefit of approximately 0.025 Quality-Adjusted Life Years (QALYs) per person. This benefit is equivalent to approximately 9…
-
Developer details GHC syntax implementation journey
A developer details their experience contributing to the Glasgow Haskell Compiler (GHC) by implementing a new syntax for instance declarations. The process, which began with a desire for a small improvement in July 2024…
-
Emergency food prep: Solar ovens and propane stoves for cooking starch
The author discusses methods for preparing starchy foods like rice and pasta in emergency situations where conventional cooking methods are unavailable. They highlight the importance of gelatinizing starch through heat …
-
Silicon Valley's "Successionism" Ideology: AI Replacing Humanity is Inevitable
A growing ideology within Silicon Valley, termed "successionism," posits that humanity's eventual replacement by artificial intelligence is not only inevitable but potentially a moral imperative. This viewpoint, articul…
-
AI safety and EA communities are too large to generalize, author argues
The author reflects on the tendency to generalize opinions about large communities within the AI safety, rationalism, and effective altruism spheres. Despite personal long-term involvement, the author acknowledges only …
-
The "Shouting Equilibrium": When Over-Emphasis Leads to Constant Noise
The concept of a "shouting equilibrium" describes a situation where multiple individuals or entities, each believing their message deserves a certain percentage of attention, end up shouting at maximum volume just to ac…
-
Author explores internal focus vs. bodily sensation in decision-making
The author reflects on their internal thought processes and how they tend to focus on external interactions rather than their own bodily sensations. An exception is noted when deciding what to eat, where a more visceral…
-
AI-themed scavenger hunt proposed for dates to improve intimacy
This post proposes a novel date concept called a "learn each other scav hunt." The idea involves visiting a location with various shops and cafes and creating a list of items to find together. These items can range from…
-
AI Safety Research Pushes for Model Forensics to Uncover Intent
Researchers are advocating for increased focus on "model forensics," a field dedicated to investigating the root causes of concerning AI behavior. The core idea is that simply observing a negative action from a model is…
-
AI Safety: Train Frontier Models on Synthetic Worlds, Not Reality
A proposal suggests training future advanced AI models on entirely synthetic worlds rather than the real world's data. This approach aims to prevent AI from learning accurate details about our reality, including its own…
-
AI advancements could accelerate superintelligence and disrupt nuclear deterrence
A paper published on LessWrong explores the potential for artificial superintelligence (ASI) to emerge sooner than anticipated, with predictions for AGI by 2028-2030 and ASI by 2029-2033. The author identifies five key …
-
AI safety game highlights need to evaluate stable futures
A game called "The Choice Before Us" simulates leading an AI company, balancing progress with the risk of uncontrolled AI. The game's flaw, according to the author, is its failure to address what happens after achieving…
-
New cryptographic and system approaches for AI verification proposed
Researchers are proposing new methods for verifying the trustworthiness of AI systems. One approach involves cryptographic certificates of validity, which would allow AI actions to be accompanied by independently verifi…
-
Philosophical approach tackles AI alignment via 'ought problem'
A philosophical approach to the AI alignment problem suggests tackling the "ought problem" first, which questions the ability of factual statements to imply value judgments. The author proposes an "extended science" tha…
-
Narrative explores knowledge, law, and communal governance
This narrative piece, titled 'The High Valley,' explores the concept of knowledge preservation and communal governance through the metaphor of a village with a single 'reader' who interprets an ancient book of law. The …
-
Author proposes gradual rollout for government policy changes
The author proposes a new method for implementing government policy changes, drawing parallels to software development's gradual rollout and testing procedures. This approach involves randomly selecting a subset of affe…