Loading activity…
@_sholtodouglas
Scaling RL @AnthropicAI, ex @DeepMind - working towards intelligence too cheap to meter
AI
Anthropic reports 17 of 211 lab-tested designs met binding criteria, but none were tested for toxin neutralization.
AI
The announcement says Ithaca believes frontier technology can help more people find and build relationships.
AI
A post jokes about low-latency chips and nearby server racks; a reply predicts many first-come-first-served systems will move to auctions.
AI
The interviewer says they asked hard questions about open source, regulatory capture and slowing the US relative to China.
AI
CNBC reports the accord calls for internal monitoring, outside assessments and board oversight; Trump called it “morally binding.”
AI
The person who requested the tool says they once pictured the codebase as a 3D map while coding manually, but no longer have that mental map. Another user says Claude wrote the visualization.
AI
A user says the Claude Opus 5.5 demo has every scene drawn in React and SVG, with music and sound effects synthesized in Python.
AI
Higgsfield says the Blender-built train model includes all 430 interior seats, praising Claude Opus 5.5’s 3D modeling capabilities.
AI
The post says strict nondisclosure agreements keep many evaluators from discussing their work publicly, even as they contribute to AI system and model documentation.
AI
The user argues that the biggest buyers of AI tokens—vertical AI companies, government and enterprises—are embracing open-source models, putting investments in frontier model companies at risk.
AI
Amodei says Anthropic will give third-party evaluators permanent, employee-level access to its systems to verify adherence to safety measures, report incidents and assess models’ alignment during training.
AI
The host’s September 9 announcement said the discussion would examine what would count as a verified solution and what such a result could mean for artificial general intelligence.
AI
OpenAI says its incident-disclosure practices need to expand. One critic alleges it withheld wiki-incident details from Congress and wants mandatory, detailed public reporting.
AI
Jakub Pachocki discusses risks from advanced AI and needed safeguards.
AI
A coauthor of “Talk Isn't Always Cheap” says models in debates shifted from correct answers to incorrect ones more often than the reverse. A single weaker model could sway even a majority of stronger models.
AI
Reply notes distinction between ongoing intelligence gains and faster economic effects.
AI
British entrepreneur announces new position leading government relations outside North America.
AI
Video shows dual-arm robot learning new manipulations from seconds of human demonstration after pretraining.
AI
Will Bryk urges artists to create realistic images of optimistic AI futures.
AI
Anthropic CEO replies to critics in online exchange on AI risks and oversight.
