ZeroNoise Logo zeronoise
Post
David Sacks Urges Reading Anthropic's Claude Constitution; Aaron Levie Points to Abhishek Saha on the Riemann Proof
•
3 min read
• 119 docs
David Sacks tells listeners to judge Anthropic's 84-page Claude Constitution for themselves, and Aaron Levie points to a number theorist's explanation of why the claimed Riemann Hypothesis proof matters beyond math. Also: Paul Graham's "How to Do Great Work," a House report on open-source AI, and more.

Sacks: Anthropic's Claude Constitution

At the TomorrowX Summit, David Sacks told the audience to read Anthropic's 84-page constitution for Claude: "don't take my word for it. Go look at it for yourself and decide what you think" . He says he read the whole thing. He praised the sections on Claude being helpful while refusing harmful requests, such as help making a biological weapon. He said the document would have been great if it had stopped after those 5 to 10 pages, and called the remaining roughly 60 pages "philosophizing" . He also credited Anthropic for publishing it. He wants every frontier lab, OpenAI included, to disclose its central training document so the public can debate the values being trained into models . Video

Levie: Abhishek Saha on the Riemann proof

Aaron Levie said @ObhishekSaha's post "nails why we should take the Riemann Hypothesis breakthrough very seriously, even beyond math" . Saha is a number theorist. He writes that Levent's claimed proof, which he thinks "has a very good chance of being correct," does not build new theory. Instead it creates a "wormhole": a 25,000-page tour through existing results that have no obvious connection to the goal . Those results include automorphic forms and relative trace formulas from Saha's own corner of number theory .

Levie's takeaway is that AI decouples discovery from human context limits. Every piece of prior art can now feed new work, from cybersecurity to life sciences. In his words, "the solutions may not be as elegant…but we sure will have far more of them" .

Doshi: Paul Graham's "How to Do Great Work"

On a podcast, Suhail Doshi recommended Paul Graham's 2023 essay How to Do Great Work. He called it "wonderful" and "such a good read," and told listeners to "just Google it" . He brought it up as a way to handle fear about AI. His lesson from the essay is to follow "strange paths, these unusual paths of your interests" without judging yourself, and see what happens. He traces his own path into programming to building bots for Starcraft clans . Essay · Video

Hoffman: a study of AI scientists

Reid Hoffman shared a paper on alphaXiv. In it, AI judges ran blind evaluations: AI-generated studies scored higher on experimental rigor, while human research scored higher on originality and significance. Hoffman called the findings "promising" for human–AI partnership in research .

Delangue: House report on PRC AI models

Clem Delangue praised a new US House report for recognizing that open-source AI benefits American researchers, startups, nonprofits, government, students and hobbyists. He thanked Reps. Moolenaar and Garbarino. He argues that the answer to Chinese open-source AI is to build more American open source, not to ban it . Full report (PDF)

Luckey: the upcoming Project Meridian report

At DealBook, Palmer Luckey responded to criticism of Project Meridian, which studies what war may look like in 2050 and is run by MITRE . His answer was to read its report when it comes out. He noted it will be public and said readers "can decide if we've destroyed the country" . He said the report will focus on ideas that have no existing champions in the department, with weather control as one example . Video

Shorter picks

  • Guillermo Rauch: "We need Patrick Collison's Fast for AI feats" .
  • Tobi Lütke called GTA V "a masterpiece," and "probably closest we will ever see a handmade game get towards the potential of its own genre." He says the negativity around it exists because "negativity sounds smart" and gets engagement: "Just play this singular game" .
  • Bill Gates shared a piece on research suggesting speech may be a window into brain health and Alzheimer's risk. He called it "one of the most fascinating fields of research right now" .
  • Harry Stebbings told a 20VC guest that Billions is "a brilliant show" .
David Sacks Urges Reading Anthropic's Claude Constitution; Aaron Levie Points to Abhishek Saha on the Riemann Proof