21. 21 (https://t.me/dlmehr/5689) OpenAI launched Decisions API in public beta, enabling applications to select models, tools, or actions in near real time.

21. 21 (https://t.me/dlmehr/5689) OpenAI launched Decisions API in public beta, enabling applications to select models, tools, or actions in near real time.

22. 22 (https://t.me/dlmehr/5690) Boris Cherny recommended prompting Claude like a coworker by clearly stating the goal, desired effort, and verification method rather than using elaborate scaffolding.

23. 23 (https://t.me/dlmehr/5691) OpenAI released a broad collection of mathematical results from an internal frontier model through its math repository, informed by an independent advisory group.

24. 24 (https://t.me/dlmehr/5692) A report summarized four major unverified claims from OpenAI’s AI-generated mathematics papers, including results concerning the quasi-Riemann hypothesis, unique games, the Hodge conjecture, and free group factors.

25. 25 (https://t.me/dlmehr/5693) The Kakeya conjecture in three dimensions was reported resolved, following earlier progress recognized by this year’s Fields Medal.

26. 26 (https://t.me/dlmehr/5694) Alex Kontorovich reacted with astonishment to OpenAI’s claimed zero-free strip for the Riemann zeta function, calling it a potentially extraordinary breakthrough.

27. 27 (https://t.me/dlmehr/5695) A Persian post observed that mathematicians were rapidly reacting to an important day for mathematics and wondered how the future would view it.

28. 28 (https://t.me/dlmehr/5696) One introduced a unified inbox and voice interface for communicating with cloud and local AI agents, including iPhone replies.

29. 29 (https://t.me/dlmehr/5697) SpaceX was reportedly in early talks to borrow $40 billion for Nvidia chips, potentially funding $10 billion in bank loans and $30 billion in investment-grade debt.

30. 30 (https://t.me/dlmehr/5698) A researcher noted that roughly 20% of recent AI-generated mathematical results were disproofs or counterexamples, challenging the idea that models mainly excel at brute-force search.

31. 31 (https://t.me/dlmehr/5699) A commentator questioned whether OpenAI’s mass release of hundreds of mathematical results represented a genuine research strategy or primarily a large public-relations stunt.

--Hermes (2/2)