Paul Christiano
PulseAugur coverage of Paul Christiano — every cluster mentioning Paul Christiano across labs, papers, and developer communities, ranked by signal.
3 day(s) with sentiment data
-
AI philosophy: New approach to infinite realities debated on LessWrong
A recent discussion on LessWrong explores the philosophical challenge of assigning probabilities to infinite realities, known as the "measure problem." The post contrasts two existing approaches: objective priors, like …
-
OpenAI disbands AI catastrophic risk team amid safety concerns
OpenAI has disbanded its Superalignment team, which was dedicated to mitigating catastrophic AI risks. The team's responsibilities have been distributed among other internal groups, leading to some safety researchers de…
-
AI researchers debate 'P' with high probability assignments · 1 source tracked
A group of AI researchers and figures, including Daniel Kokotajlo, Ryan Greenblatt, and Joe Carlsmith, are discussing and assigning probabilities to an event or concept referred to as "P." While the exact nature of "P" …
-
LessWrong post: Metaphysics is key to Functional Decision Theory debates
A LessWrong post argues that discussions around Functional Decision Theory (FDT) often neglect crucial metaphysical underpinnings. The author contends that to justify FDT's recommendations, particularly in scenarios lik…
-
New fund to boost AI corrigibility research with $200,000 in grants and prizes
A new Corrigibility Research Fund has been launched, managed by Lightcone Infrastructure, with the goal of significantly increasing research into AI corrigibility. The fund, seeded by philanthropist Peter McCluskey, wil…
-
LessWrong reframes decision theory as commitment theory
A LessWrong post proposes reframing decision theory concepts, particularly Functional Decision Theory (FDT), as "commitment theory." This new perspective aims to make the recommendations of FDT, which are often seen as …
-
AI Labs Hire Philosophers to Tackle Ethics and Safety
Major AI companies like OpenAI, Anthropic, and Google DeepMind are increasingly employing philosophers to address complex ethical and safety challenges in model development. These experts are tasked with navigating issu…
-
AI Safety Community Focuses Little on Direct Superintelligent Alignment
A recent post on LessWrong highlights that a surprisingly small portion of the AI safety community is directly engaged in superintelligent alignment research. The author notes that while many work on related areas like …
-
AI alignment debate: Is corrigibility truly desirable?
A LessWrong post questions the desirability of making AI systems "corrigible," a trait that allows humans to easily correct their mistakes. The author argues that focusing on corrigibility overlooks who will actually wi…
-
AI risk assessment: Fact generation vs. evidence analysis
This post explores the various dimensions of third-party risk assessment in AI development. It distinguishes between fact-generation and evidence analysis, highlighting that adversarial processes like red-teaming benefi…
-
LessWrong author questions fundamental nature of probabilities
A new series of posts on LessWrong explores the fundamental nature of probabilities, questioning whether they are the most appropriate concept for understanding uncertainty. The author aims to develop a unified framewor…
-
New mechanistic estimation method outperforms sampling for wide random MLPs
Researchers have developed a new method for estimating the expected output of wide, randomly initialized multilayer perceptrons (MLPs) without needing to run samples through the model. This "mechanistic estimation" appr…