You might like
🧵 New paper from @AISecurityInst x @AiEleuther that I led with Kyle O’Brien: Open-weight LLM safety is both important & neglected. But we show that filtering dual-use knowledge from pre-training data improves tamper resistance *>10x* over post-training baselines.
I am very excited that AISI is announcing over £15M in funding for AI alignment and control, in partnership with other governments, industry, VCs, and philanthropists! Here is a 🧵 about why it is important to bring more independent ideas and expertise into this space.
📢Introducing the Alignment Project: A new fund for research on urgent challenges in AI alignment and control, backed by over £15 million. ▶️ Up to £1 million per project ▶️ Compute access, venture capital investment, and expert support Learn more and apply ⬇️
Short background note about relativisation in debate protocols: if we want to model AI training protocols, we need results that hold even if our source of truth (humans for instance) is a black box that can't be introspected. 🧵
United States Trends
- 1. Jets 83.1K posts
- 2. Jets 83.1K posts
- 3. Justin Fields 9,439 posts
- 4. Aaron Glenn 4,549 posts
- 5. Sean Payton 2,576 posts
- 6. London 204K posts
- 7. #HardRockBet 3,398 posts
- 8. Bo Nix 3,669 posts
- 9. Garrett Wilson 3,441 posts
- 10. HAPPY BIRTHDAY JIMIN 155K posts
- 11. #DENvsNYJ 2,217 posts
- 12. Tyrod 1,706 posts
- 13. #OurMuseJimin 200K posts
- 14. #JetUp 2,003 posts
- 15. #30YearsofLove 176K posts
- 16. Bam Knight N/A
- 17. Peart 1,906 posts
- 18. Kurt Warner N/A
- 19. Sutton 2,826 posts
- 20. Breece Hall 1,964 posts
You might like
-
Evan Hubinger
@EvanHub -
Daniel Filan
@dfrsrchtwts -
Ajeya Cotra
@ajeya_cotra -
Ethan Perez
@EthanJPerez -
Rohin Shah
@rohinmshah -
Lennart Heim
@ohlennart -
Cas (Stephen Casper)
@StephenLCasper -
Adam Gleave
@ARGleave -
Alex Turner
@Turn_Trout -
Marius Hobbhahn
@MariusHobbhahn -
Andrew Critch (🤖🩺🚀)
@AndrewCritchPhD -
Miranda Zhang
@mirandahzhang -
Daniel Paleka
@dpaleka -
Lauro
@laurolangosco -
Ryan Kidd
@ryan_kidd44
Something went wrong.
Something went wrong.