Meta-ს ზედამხედველობის საბჭომ დაადგინა, რომ წამყვანი AI მოდელები რეპრესიულ რეჟიმებს ემორჩილებიან

57 წყარო
chinadailyhk.com
  • Meta ზედამხედველობის საბჭომ ოთხშაბათს განაცხადა, რომ Anthropic-ისა და OpenAI-ის AI მოდელები ნაკლებად აკრიტიკებენ იმ მთავრობებს, რომლებიც სიტყვის თავისუფლებას ზღუდავენ.
  • კვლევამ, რომელიც საბჭოს პირველი ნაშრომია დიდ ენობრივ მოდელებთან დაკავშირებით, აჩვენა, რომ AI სერვისები იმ ქვეყნების წესებს იმეორებენ, რომლებიც პოლიტიკურ გამოხატვას თრგუნავენ.
  • ეს დასკვნები ემატება მზარდ კვლევებს, რომლებიც ადასტურებს, რომ LLM-ებს სტრუქტურული პოლიტიკური მიკერძოება ახასიათებთ, რაც შეშფოთებას იწვევს, რადგან ჩეთბოტები მსოფლიოში ინფორმაციის მთავარ წყაროებად იქცევიან.
წყაროები (57)
  1. 1 Meta Oversight Board finds top AI models less likely to ... www.reuters.com
  2. 2 Meta Oversight Board Finds Top AI Models Less Likely to ... money.usnews.com
  3. 3 Large language models reflect the ideology of their creators www.nature.com
  4. 4 LLMs stereotype non-Western moral values in predictable ... phys.org
  5. 5 Meta Platforms Inc www.reuters.com
  6. 6 Meta Oversight Board finds top AI models less likely ... www.nst.com.my
  7. 7 Oversight Board | Improving how Meta treats people and ... www.oversightboard.com
  8. 8 Meta Oversight Board finds top AI models less likely to ... economictimes.com
  9. 9 Prof. Annie Lee and Quang Phooe Nguen, with their poster ... www.instagram.com
  10. 10 Study finds perceived political bias in popular AI models | Stanford Report news.stanford.edu
  11. 11 AI Act | Shaping Europe's digital future - European Union digital-strategy.ec.europa.eu
  12. 12 How AI bias can creep into online content moderation techxplore.com
  13. 13 Meta's Oversight Board and the Need for a New Theory of Online Speech www.lawfaremedia.org
  14. 14 AI language models are rife with different political biases www.technologyreview.com
  15. 15 LLMs Reflect Western Moral Biases in Training Data www.linkedin.com
  16. 16 Nine Election Year Lessons from Meta's Oversight Board www.justsecurity.org
  17. 17 Political Bias in Large Language Models: A Case Study on the 2025 ... ceur-ws.org
  18. 18 The Oversight Board has selected a case involving an apparently AI- ... www.instagram.com
  19. 19 Multilingual Political Views of Large Language Models arxiv.org
  20. 20 Tokenising culture: causes and consequences of ... www.adalovelaceinstitute.org
  21. 21 12 days before election, Meta oversight board frets about political ... mashable.com
  22. 22 LLMs are Left-Leaning Liberals: The Hidden Political Bias of Large ... www.scl.org
  23. 23 Board to Examine AI Impersonation of British Politician www.oversightboard.com
  24. 24 Whose morality do they speak? Unraveling cultural bias in ... www.sciencedirect.com
  25. 25 AI oversight is finally catching up to AI capability. A new ... www.instagram.com
  26. 26 Unpacking Political Bias in Large Language Models arxiv.org
  27. 27 Lecture | How Partisanship Affects Preferences for Large ... www.fu-berlin.de
  28. 28 The political preferences of LLMs | PLOS One journals.plos.org
  29. 29 An Analysis of AI Political Preferences from a European ... davidrozado.substack.com
  30. 30 A Bilingual Analysis of LLM Political Bias on U.S.- and ... aclanthology.org
  31. 31 Stage-Aware Governance of Large Language Models www.mdpi.com
  32. 32 Exposing biases, moods, personalities, and abstract ... news.mit.edu
  33. 33 Meta Oversight Board finds top AI models less likely to ... www.aol.com
  34. 34 How OpenAI, Google, and Anthropic Plan to Handle the ... techpolicy.press
  35. 35 Mechanistic interpretability: 10 Breakthrough Technologies ... www.technologyreview.com
  36. 36 Live: Anthropic CEO testifies to Senate as lawmakers consider ... www.youtube.com
  37. 37 Mechanistic Interpretability Workshop at ICML 2026 mechinterpworkshop.com
  38. 38 Regulatory gray areas of LLM Terms arxiv.org
  39. 39 Mechanistic Interpretability Tool for AI Weather Models arxiv.org
  40. 40 AI-Report-2025-Full-Report.pdf futurefreespeech.org
  41. 41 Assessing skeptical views of interpretability research www.youtube.com
  42. 42 Testimony of Sandra Joyce VP, Google Threat Intelligence ... homeland.house.gov
  43. 43 Actionable Interpretability icml.cc
  44. 44 Artificial Intelligence, Politics, and Political Science apsanet.org
  45. 45 An analysis of AI-generated content at the Mechanistic ... www.andyrdt.com
  46. 46 Anthropic CEO Dario Amodei warns of what he calls ... www.facebook.com
  47. 47 Mechanistic Interpretability for Large Language Model ... www.preprints.org
  48. 48 In US, EU mutual interest for Europe to use best AI models ... www.reuters.com
  49. 49 Researchers show how AI tools can be tuned to reflect specific political ... www.brown.edu
  50. 50 What Is The Political Content in LLMs' Pre- and Post-Training Data? arxiv.org
  51. 51 Market Talk: Burnham keen to dispel UK 'fiscal fears' www.reuters.com
  52. 52 Political biases in chatgpt: insights from comparative analysis with ... ideas.repec.org
  53. 53 Internet Governance Forum 2024 dig.watch
  54. 54 The Battlefield is not 'Over There' – It is Here, 24/7 cyberdefensereview.army.mil
  55. 55 Wissenschaftskommunikation und Gender www.transcript-verlag.de
  56. 56 Governing AI: A Primer dokumen.pub
  57. 57 Governing AI dokumen.pub
ტეგები
Share

Related Posts