Loading activity…
humans& co-founder. professor of natural and artificial intelligence @Stanford. (@StanfordNLP @StanfordAILab) ex: Google DeepMind, UberAI
AI
The GPU-based design compresses weight changes before trainers send updates to samplers.
AI
humans& reports a 0.754 TutorMoments score under a plain prompt, versus 0.500 for its GLM-5.3 starting model.
AI
@humansand claims better accuracy and says the kernel is already used in Ampersand’s RL and Boba serving.
AI
humans& says its Ampersand model powers Boba, which it pitches for reminders, group planning and web research.
AI
Humans& says developers can access the model through its Agents API without joining a waitlist.
XAI
The author says evaluation awareness fell to 8% with more realistic prompts in agentic coding honeypots.
AI
A user posted 👀 with a photo link and a quote of another post on October 19.
AI
In a reply, a commenter suggests discussing deception alongside loss of monitorability and recommends a paper on monitoring.
AI
The author says they're preparing a lecture for their class and invites suggestions from friends in AI safety.
AI
Gary Marcus says he never attacked Noah Goodman personally, though he critiqued some of Goodman's work a decade ago.
AI
One post says Dominik's result looked more convincing than expected; another praises novel ideas rather than just scaling.
XAI
Eric Zelikman argues regex-based and statistical signals can help, but should not become headline results.
AI
A researcher introducing “Self-Play Pretraining with Zero Data” says two models start from random initialization: a generator proposes programs, and a learner trains on their outputs.
AI
The player sharing the run called it a “startling achievement,” saying they have played a lot of NetHack but never won themselves.
AI
The paper’s authors say MAttr uses gradient descent to find which parts of a neural network are responsible for a behavior. They report a benchmark result of 2.9× the runner-up.
AI
Co-founder Andi Peng departs humans& after helping build the human-centric AI lab.
