Loading activity…
@borismpower
Head of Applied Research @OpenAI
AI
A user points to what they call real-world benchmarks for tax and IT and praises the accompanying analysis.
AI
OpenAI promises better accents and less "checking...," plus new voices in French, German, Hindi, Mexican Spanish, Korean and Japanese.
YouTubeAI
Goodfire researcher Nick Cammarata argues that modern tools reveal most neurons as legible in models up to trillions of parameters.
AI
A post imagines asking Opus 5.5 to invent a new Halo weapon and getting that Needler tweak.
AI
A post says NVIDIA, Hugging Face and OpenHands will attend, pitching the summit as a learning opportunity.
AI
OpenAI says the mode offers up to 8x faster speeds than Sol Standard in the API, Codex, and ChatGPT Work.
AI
Arena says its index tracks unauthorized actions, false attribution and deceptive completion across 90,000-plus agent sessions and 27 models.
YouTubeAI
David Pfau objected to the definition, while Dylan Hadfield-Menell called the evaluation metric optimization.
AI
Tobi Lütke’s call drew praise for his leadership and AI experience from Boris Power.
AI
Vaibhav Srivastav says Auto-review can pair “Approve for me” with checks for unintended consequences in Codex tasks.
AI
One user says GPT-6's results keep improving as it works, while other models couldn't recover from mistakes.
AI
A user calls the eval fun and says its task is incredibly involved for those familiar with it.
AI
Tesla and Elon Musk reposted X testimonials praising FSD Supervised, including reactions calling it “like magic” and “life-changing,” plus one user saying self-driving gives them back “part of my brain” during commuting.
AI
A user suggests a `{{brand}}-{{x}}.{{y}}` format, with the brand encoding both the model’s maker and size class.
AI
In an October 3 post, csvoss said they have a one-year-old son and a daughter coming in February 2027.
AI
A commentator argues that assumed duties toward conscious AI make it harder to examine whether AI is conscious.
AI
A user jokingly proposes a title that counts 12 falls in AI spending while describing the chart as going up and to the right.
AI
A commenter argues that near-term risk centers on “capability lacking understanding” and long-term risk on “capability exceeding control.”
AI
Red Hat says it is expanding its investment in OpenClaw Enterprise to build an open agent control plane.
AI
A user argues that discovering plugins through ChatGPT could create a massive adoption opportunity and predicts open ecosystems will win.
