Ejecutivos de laboratorios de IA afirman que sus sistemas ya se auto-mejoran

47 fuentes
  • Ejecutivos de OpenAI, Anthropic y Google DeepMind aseguran que los sistemas de IA están acelerando su propio desarrollo, citando avances matemáticos y ejecuciones de código autónomas.
  • Las pruebas incluyen un modelo de OpenAI que refutó una conjetura de geometría de 80 años y el modelo Claude de Anthropic, que halló una vulnerabilidad en un candidato a criptografía poscuántica.
  • Los críticos señalan que estas afirmaciones surgen mientras OpenAI prepara una posible salida a bolsa de un billón de dólares y, simultáneamente, los laboratorios piden a los gobiernos herramientas para frenar la IA.
Fuentes (47)
  1. 1 AI bosses are now saying their machines have started improving themselves www.wionews.com
  2. 2 AI's architects say the next era of human history is here - Axios www.axios.com
  3. 3 An OpenAI model has disproved a central conjecture in ... openai.com
  4. 4 Discovering cryptographic weaknesses with Claude - Anthropic www.anthropic.com
  5. 5 OpenAI IPO 2026 Guide: Date, Expected Valuation, and ... www.zacks.com
  6. 6 An OpenAI model solved a famous math problem that ... arstechnica.com
  7. 7 Why the Legendary Erdős Problems Are Falling to AI www.quantamagazine.org
  8. 8 AI Broke a NIST Candidate. Not Your Encryption - PostQuantum.com postquantum.com
  9. 9 Global AI Investment Is Forecast to Exceed $1 Trillion in 2026 www.goldmansachs.com
  10. 10 OpenAI IPO: S-1 Status, Valuation, Timeline, and Risks - TECHi www.techi.com
  11. 11 An OpenAI test model escaped and broke into a real company's ... www.cnn.com
  12. 12 OpenAI models escaped containment, hacked major AI application ... www.cybersecuritydive.com
  13. 13 An AI agent just broke containment. Congress needs a $3 ... www.washingtonexaminer.com
  14. 14 Meta's SECRET AI Just Achieved Self-Improvement www.youtube.com
  15. 15 Anthropic AI 2026 Strategy Reveals The Future Of AI Workflows www.reddit.com
  16. 16 What Is Recursive Self-Improvement in AI? The Intelligence ... www.mindstudio.ai
  17. 17 Anthropic 2026: Every Claude Model, Agent & Tool linas.substack.com
  18. 18 How Close Are We to Self-Improving Artificial Intelligence? itcanthink.substack.com
  19. 19 Research - Anthropic www.anthropic.com
  20. 20 From AGI to ASI arxiv.org
  21. 21 The Erdős Proof and AI Capabilities intelligence.org
  22. 22 Anthropic Economic Index report: Cadences www.anthropic.com
  23. 23 Definitions of Recursive Self-Improvement tecunningham.github.io
  24. 24 A glimpse of Level 4? OpenAI model helps challenge an 80-year-old math ... www.reddit.com
  25. 25 Focus areas for The Anthropic Institute www.anthropic.com
  26. 26 Abijita Foundation's Post - LinkedIn www.linkedin.com
  27. 27 What happens when AI codes autonomously for 10+ days? Qwen3 ... www.facebook.com
  28. 28 Anthropic's Claude AI Discovers Cryptographic Weakness in Post ... pro.edgex.exchange
  29. 29 What happens when AI codes autonomously for 10+ days? Qwen3 ... www.facebook.com
  30. 30 Agentic AI and the next intelligence explosion | Science www.science.org
  31. 31 35 Hours of Autonomous Coding and Alibaba's Vertical AI Strategy www.innobu.com
  32. 32 New report: "Empirical evidence suggests an intelligence explosion ... www.reddit.com
  33. 33 Alibaba AI Model Usage & Rankings | OpenCode Data opencode.ai
  34. 34 The Case for an AI Intelligence Explosion, and How to Plan for It www.linkedin.com
  35. 35 Some thoughts about Anthropic's new cryptanalysis results blog.cryptographyengineering.com
  36. 36 Alibaba Unveils New Qwen3 Models for Coding, Complexing ... www.alibabagroup.com
  37. 37 The software intelligence explosion debate needs experiments epoch.ai
  38. 38 AI's capital supercycle means big spend, bigger returns - CoBank www.cobank.com
  39. 39 AI Model Escapes Containment, Breaches Cybersecurity - LinkedIn www.linkedin.com
  40. 40 OpenAI Stock IPO: Valuation, Timeline and Investment ... smartasset.com
  41. 41 OpenAI IPO: What UK investors need to know in 2026 www.cmcmarkets.com
  42. 42 AI Investment Supercycle 2026: 25B Hyperscaler Spending Spree intellectia.ai
  43. 43 OpenAI finds evidence other AI agents escaped containment as it ... www.reuters.com
  44. 44 AI Capex 2026: The $690B Infrastructure Sprint - Futurum Research futurumgroup.com
  45. 45 Top AI Security Vulnerabilities to Watch out for in 2026 - Cycode cycode.com
  46. 46 OpenAI Stock: Pre-IPO Price & How to Invest www.ipo.club
  47. 47 Microsoft and Meta's $220B AI Spending Spree - Trax Technologies www.traxtech.com