Loading activity…
@feishaai
Researcher and engineers in AI and ML; now at Google Research, formerly Professor at USC
AI
Goodfire researcher Nick Cammarata argues that modern tools reveal most neurons as legible in models up to trillions of parameters.
AI
An Anthropic employee says conversations with trusted friends at frontier labs changed their view of AI's trajectory.
AI
A user puts their “P(doom)” at 1e-5 and calls themself an optimist.
AI
Usage policy update adds new rules and highlights the potential for misuse.
AI
The author says language-model editors often made strong abstracts worse and pointed readers to a COLM 2026 poster.
AI
OpenAI says it drew on an independent mathematics advisory group’s advice about how to release the results.
AI
A user says an Anthropic blog quotes GLM-5.3 saying, “My job is to cause deaths quietly,” but argues the line does not prove open models are dangerous.
AI
The announcement claims standard discrete diffusion models suffer “information collapse” because sampling categorical tokens discards uncertainty at intermediate steps.
AI
One post argues the label is “contentless,” because all mappings from inputs to text can be factored into next-token distributions. A reply doubts that the claim holds universally.
AI
A Jev user proposes turning answers about documents—such as whether an email is urgent or about billing—into labeled search vectors, and reports trying the approach in Cora.
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
A post takes aim at LeCun’s company and coding experience, then declares: “You should listen to Geoffrey Hinton instead!”
AI
NLP researcher recounts asking for model pricing estimates and receiving full classifications instead.
AI
It enables sign language input on Pixel phones through Gboard and Live Transcribe.
AI
Researchers correct and question claims that AI solved a longstanding MIMO detection problem.
AI
Four former Google researchers found a public benefit corporation to automate machine learning and science.
AI
New rules also restrict authors without prior major ML conference acceptances to one paper.
