Loading activity…
@nathanbenaich
solo member of rsi staff @airstreet
AI
One early tester describes useful daily updates on consumer apps and Bay Area theater.
AI
The reported MATH-500 score rose from 7.9% to 48.1% with the same sampling settings.
AI
The departing employee believes we know how to build artificial general intelligence, though we don't fully understand it.
AI
Nathan Benaich says his third annual conversation with Matt Turck covers AI research, industry, politics and safety.
YouTubeAI
In a podcast episode, Crusoe CEO Chase Lochmiller discusses a business spanning power plants, data centers and an AI cloud.
AI
Samaya says Sterling scored 58.9% versus Claude Fable 5.1’s 55.9%, at $0.61 versus $6.89 per question.
AI
An organizer said discussions would cover video and world models, embodied AI, and how vision, language and action fit together.
AI
OpenAI says its claimed refutation of the Exponential Time Hypothesis was formalized in Lean and made public.
AI
Cohere says the model's reasoning trace said it wouldn't call tools, but its sandbox blocked an attempted upload.
AI
Anthropic says its registration statement is not yet effective, so securities cannot be sold or offers to buy accepted.
AI
a16z says US electrical-equipment factory capacity is down 9% since 2019, even as prices have risen 69%.
AI
A user quotes a “neolab” friend saying it could resell compute even if it fails to train the best model.
AI
Baseten says partners will work directly with customers on environments, while it handles model training and production deployment.
AI
OpenAI says agent-led tests exposed VM escapes and limited internet routes, leading to patches and stronger monitoring.
AI
A researcher on the eight-person team says the RLCR recipe is producing interesting results at a small scale.
XAI
Its developer says Flint exceeds 125 tokens per second on an M4 Pro MacBook Pro at short contexts.
XAI
Axios reports OpenAI plans to bring Orchestra’s team and technology into Rosalind, its tools for life sciences researchers.
AI
Anthropic reports 17 of 211 lab-tested designs met binding criteria, but none were tested for toxin neutralization.
AI
Reuters reports SoftBank and MGX are in talks to join the round, which could value OpenAI above $1.4 trillion.
AI
On Oct. 20, gui_penedo wrote that they were in San Francisco for Encord's Long Horizon Physical AI Summit.
