Loading activity…
@tdietterich
University Distinguished Professor (Emeritus), Oregon State Univ.; Former President, AAAI; ArXiv Chair CS Section, Chair Editorial Advisory Council
AI
A post argues open-weight cyber models benefit defenders, citing progress finding bugs and the cost of comparable closed models.
AI
A user praises the claimed restriction and argues that every country should adopt it.
AI
Goodfire researcher Nick Cammarata argues that modern tools reveal most neurons as legible in models up to trillions of parameters.
AI
A user argues that today’s most capable AI operates like human society, with intelligence rooted in cooperation.
AI
A user asks what remains mysterious about generalization; a reply points to defining “domain” and distinguishing generalization from memorization.
AI
A poster says almost any Stage 4 cancer patient would find human understanding of a working cure's mechanism irrelevant.
AI
Satya Nadella argues organizations should control what models can access and do, with safeguards outside the models themselves.
AI
Usage policy update adds new rules and highlights the potential for misuse.
AI
Dillon Plunkett argues the reported axis may reflect language representations, not pain experienced by a model.
AI
OpenAI says it drew on an independent mathematics advisory group’s advice about how to release the results.
AI
arXiv says the change aims to distribute moderator time fairly and support staff, volunteers, readers and authors.
AI
A commentator challenges the claim that large language models are limited to memory and retrieval, arguing that their learned representations allow them to find new solutions.
AI
A user is puzzled by reasoning traces they consider remarkably good, questioning how reinforcement learning could produce them—even with initial chain-of-thought abilities and billions of rollouts.
AI
An organizer says the September 6 gathering included talks by Chris Hitchcock and Daphne Koller, plus a panel and personal reflections from Pearl’s students, collaborators and friends.
AI
Citing Nancy Leveson’s Engineering a Safer World, a thread frames safety as something that requires active control—not an inherent property of an AI agent or self-driving car.
AI
A user calls V4.1 “very brittle,” claiming it can see “massive gains” when AGENTS.md instructions are optimized for a particular scenario.
AI
Cognitive scientist argues trustworthiness, not oversight, is the fundamental AI problem.
AI
Machine learning professor replied to discussion on controlling rogue AI agents.
AI
Stanford researcher replies that the term misleads by suggesting ordinary technology.
AI
Discusses scenarios where AIs escape labs, hobbyists, or gain legal rights.
