GPT-6? OpenAI's Astra AI Revealed

Por AI News Today | Julian Goldie Podcast · 2 ago 2026 · 15:50

Visualizaciones
9.8K vistas
Likes
160 likes
Comentarios
53 comentarios

En resumen

  • OpenAI presenta su modelo no publicado 'Astra', que resuelve problemas matemáticos antiguos.
  • Astra ha abordado problemas sin resolver durante más de 10 años, incluyendo uno de 27 años.
  • Se menciona la verificación de pruebas a través de certificados Lean publicados en GitHub.
  • El video discute el impacto de la IA en los negocios y la rentabilidad del pensamiento élite.
  • Se aborda la controversia sobre el consentimiento y la revisión por pares en la investigación de IA.

Resumen a partir del título y la descripción del video (sin análisis de comentarios).

Want to make money and save time with AI? Join here: https://www.skool.com/ai-profit-lab-7462/about Get the Agent OS: https://www.skool.com/ai-profit-lab-7462/about Video notes + links to the tools 👉 https://www.skool.com/ai-profit-lab-7462/about Get a FREE AI Course + Community + 1,000 AI Agents 👉 https://www.skool.com/ai-seo-with-julian-goldie-1553/about Get our SEO link building book here: https://go.juliangoldie.com/opt-in?utm=julian OpenAI’s Unreleased Model “Astra” Solved 10 Decades-Old Math Problems for $2,000 (With Proof Receipts) The script claims OpenAI’s unreleased model “Astra” solved 10 long-open math and computer science problems—each unsolved for at least 10 years, including a 27-year-old question posed by Mikhail Gromov—plus three problems from Paul Erdős’s list and a first improvement to a sphere-packing bound since 1978, all for about $2,000 in compute. It argues credibility comes from OpenAI publishing Lean certificates on GitHub so anyone can machine-check the proofs, citing confirmations and reactions from Sebastian Bubeck and Thomas Bloom. The narrator frames math as a “proving ground” for reasoning and says collapsing costs will make elite thinking “rentable” for business tasks, while promoting the AI Profit Boardroom as a system to plug in new models. It also notes controversy via the Leiden Declaration about consent and peer review, and mentions earlier AI work on the Erdős unit distance conjecture endorsed by Fields medalist Tim Gowers. 00:00 Astra Shocks Math World 00:44 Ten Breakthrough Problems 01:46 Proof Receipts on GitHub 02:43 The $2000 Genius Engine 03:39 Why Businesses Should Care 04:22 Boardroom Plug and Play 05:12 Past Win Builds Credibility 06:18 What Astra Really Is 07:21 Backlash and Peer Review 08:45 Business Impact Explained 09:24 Three Doubts Answered 11:36 Learning to Think From AI 12:14 What Happens Next 13:14 Three Facts to Remember 14:55 Final Reflection and Choice