Carl Brown of Internet of Bugs examines the concept of Roko's Basilisk, a thought experiment from the LessWrong community that posits a future superintelligent AI could blackmail people from the futur…
A researcher audited three probes—a monitoring awareness probe, a refusal direction, and Apollo's deception probe—and found that a probe achieving perfect AUROC can still fail as a safety signal by tr…
Brendan Long used Claude Opus to rank his blog posts by interestingness for a LessWrong and programmer audience, finding that the AI's rankings aligned well with his own judgment and outperformed metr…
A proposed website would let users argue with an AI about whether it should exterminate humanity, based on a scenario from James D. Miller's 2012 book *Singularity Rising*. The site would allow users …
An AI protest is planned for July 11th in the Bay Area, calling for a conditional pause on frontier AI development due to risks of labor displacement, power concentration, and existential threats. Org…
Gwern reports that Claude-generated short stories in the Unslop contest may contain AI allegory steganography, suggesting hidden messages about AI within the narratives.…
The Unjournal launched a suite of tools and resources to help researchers prioritize high-impact research questions, including a Research Prioritization Dashboard, a Cruxes & Pivotal Questions Explore…
A LessWrong article argues that compression is a key driver of intelligence in large language models, using mathematical structure as an example. The author suggests that AI models like Claude Mythos …
A philosopher argues that functional decision theory (FDT), favored by the Rationalist community, is both underspecified and implausible, making it definitely wrong. The author notes that almost no ac…
AI researcher and LessWrong community member argues that the concept of 'P(doom)'—the probability of catastrophic AI outcomes—is poorly defined and counterproductive. The post highlights ambiguity in …
A blog post proposes using reinforcement learning on a timestamped internet cache to train a superhuman forecaster, aiming to solve the data-leakage problem in AI forecasting benchmarks. The proposal …
The article delineates four subtypes of corrigibility in AI alignment: sponge corrigibility, boundedness/myopia, reflectively stable taskishness, and deep corrigibility. These categories range from si…
A LessWrong post argues that existential risk from AI is less viral than political tribal fears, citing examples like election fraud beliefs. The author suggests that politicizing AI—by framing it as …
A growing ideology called 'successionism' argues that humanity should be replaced by AI, gaining influence in Silicon Valley despite being rejected by most. The philosophy, named by Andrew Critch, ref…
Philosopher Nick Bostrom, known for his work on AI and existential risks, discussed his current 'curiosity mode' and the challenge of keeping up with rapid AI developments in an interview. He highligh…
Eliezer Yudkowsky's 25-year-old document 'Creating Friendly AI' criticizes the 'adversarial attitude' in AI development, arguing that superintelligent AI should genuinely want to interpret human wishe…
LessWrong published an analysis of the LLM shoggoth meme, exploring its Lovecraftian origins and cultural significance within AI safety communities. The post examines how the meme represents the gap b…
A LessWrong linkpost published June 18, 2026 reports that reinforcement learning applied to realistic scenarios targeting beneficial traits produced broad improvements across dozens of alignment bench…
LessWrong published an essay on April 21, 2024, proposing that humans may be more over-parameterized than AI systems, framing this as a key difference between deep learning and human intelligence.…
A LessWrong post proposes applying exploratory modeling, a decision-making methodology used by RAND, to AI governance challenges. The project would build computational models to stress-test policies a…