Robots got a new boss, OpenAI posted a mountain of math, and New York politicians heard a warning they won't forget soon. Here's your AI day.
OpenAI just published 722 math papers written by an AI model you can't use. They come from an unreleased internal model, and the GitHub repo's own count is "722 manuscripts organized into 372 families," with each family grouping related results. The Decoder's headline frames the release as "372 proofs," which counts the families, not the papers. Many include Lean proofs, which a computer can check line by line, but the repo itself says "some of the unformalized results could have issues". Mathematicians are split. In a July blog post, Fields Medal winner Timothy Gowers warned that within a decade or two the math literature could be "vastly expanded" with no community of human experts who share an understanding of it. We went through what has actually been checked in our full story.
Boston Dynamics has a new CEO, and he comes from AI rather than robotics. Rohit Prasad, who ran Alexa and artificial general intelligence work at Amazon (AMZN) and led its Nova models, starts today. He replaces interim chief Amanda McMaster, who stepped in when Robert Playter left in February. He's only the third person to run the company since it was founded in 1992. Boston Dynamics, controlled by Hyundai, says the hire moves along its "physical AI" plan to sell smart machines at scale. Its Atlas humanoid is the robot to watch.
A former Anthropic researcher gave New York City lawmakers his personal estimate of the risk from AI. At Monday's Council hearing, Jacob Coxon said it's "more likely than not that humanity loses control to these AIs" if things stay on the current path. He framed that as his own assessment and called for slowing frontier AI work until it can be made safe. Anthropic, OpenAI, Google and Meta sent executives, who testified remotely. Elon Musk's SpaceXAI skipped it despite a subpoena, and Council Speaker Julie Menin said "we are pursuing that subpoena in court". The Council is weighing 10 bills, including one requiring a way for humans to shut AI models down. Our full story has more from the hearing.
DeepSeek, the Chinese lab known for cheap AI, is pulling in very big money. Bloomberg reported that it's close to raising at least 80 billion yuan (US$12 billion), far above its original goal of about 50 billion yuan. Tencent and battery giant CATL are among the largest contributors, according to people Bloomberg spoke with. DeepSeek hasn't confirmed the round, so treat it as a report for now. It's also reportedly preparing a possible listing on Shanghai's STAR Market. More in our story.
AMD (AMD) still can't make enough AI chips. "We plan to increase supply substantially in 2027, but clearly more will be needed," CEO Lisa Su said in Taiwan. She added that AMD now plans factory capacity with partners three to five years ahead, up from one to two. Our story has the details.
So here's today's question: if an AI proves a theorem and no human fully understands the proof, does it count as math?
The WattsUpNext take (opinion): AI labs should put as much effort into checking their work as into producing it.
That's wattsup, and here's wattsnext: Jaguar reveals its $130,000 Type 01 electric GT, on sale in 2027
Sources
- 1.openai/math repository · OpenAI (GitHub)
- 2.OpenAI dumps 372 AI-generated math proofs on GitHub, telling the academic world to keep up · The Decoder
- 3.Boston Dynamics appoints former Amazon exec Rohit Prasad as new CEO · The Robot Report
- 4.New York City Council holds landmark AI oversight hearing · CBS New York
- 5.DeepSeek reportedly nears $12bn funding round backed by Tencent and CATL · The Next Web
- 6.AMD's Lisa Su Plans Chip Supply Three to Five Years Ahead as AI Demand Runs Hot · Seoul Economic Daily
- 7.Thoughts about the Leiden Declaration · Gowers's Weblog (Timothy Gowers)
- 8.DeepSeek to Raise at Least $12 Billion in Tencent-Backed Funding · Bloomberg
- 9.AI giants give few clear answers to key safety questions at NYC Council hearing amid whistleblower warnings · amNY
Reported by the WattsUpNext desk from the sources linked below. Spot an error? Tell us at corrections@wattsupnext.com.
The WattsUpNext Brief
Get stories like this in your inbox.
One email each weekday, only the topics you choose. Real news, sourced, no fluff. Unsubscribe in one click.
