aggliu

• aggliuMar 31, 2025, 6:43 PM

  1 point

  0

  in reply to: Pat Myron’s comment on: Will Je­sus Christ re­turn in an elec­tion year?

  This is probably well-known at this point, but the $6 million item is not the duct-taped banana itself, but the right to display the duct-taped banana. That may not make it any more or less sensible, though.

• aggliuMar 11, 2025, 4:38 PM

  1 point

  0

  in reply to: Gurkenglas’s comment on: Self-fulfilling mis­al­ign­ment data might be poi­son­ing our AI models

  Yep, that’s why I mentioned evil numbers specifically.

• aggliuMar 4, 2025, 6:24 PM

  LW: 6 AF: 1

  1

  AF

  on: Self-fulfilling mis­al­ign­ment data might be poi­son­ing our AI models

  I am a bit worried that making an explicit persona for the AI (e.g. using a special token) could magnify the Waluigi effect. If something (like a jailbreak or writing evil numbers) engages the AI to act as an “anti-𐀤” then we get all the bad behaviors at once in a single package. This might not outweigh the value of having the token in the first place, or it may experimentally turn out to be a negligible effect, but it seems like a failure mode to watch out for.

Can Knowl­edge Hurt You? The Dangers of In­fo­haz­ards (and Exfo­haz­ards)

S-Risks: Fates Worse Than Ex­tinc­tion

How to Upload a Mind (In Three Not-So-Easy Steps)

How to Erad­i­cate Global Ex­treme Poverty [RA video with fundraiser!]