Skip to main content
MANIFOLD
Will Yudkowsky substantially lower his P(doom) in response to Cammarata's "Neural Networks Are Already Clean"?
5
Ṁ100Ṁ82
Dec 31
25%
chance

The work in question: https://already-clean.goodfire.com/

I don't have a good source for Yudkowsky's previously reported P(doom), but it's safely 0.95 or higher. roon thinks Yud will move his P(doom): roon (@tszzl) on X

This market resolves as follows:

  • If I find out Yud was below 0.90 before the new results, resolves NA (violates the premise of the market)

  • If Yud reports a revised P(doom) below 0.70, resolves YES

  • If Yud reports a revised P(doom) between 0.70 and 0.90, resolves proportionally

  • If Yud reports a revised P(doom) or does not report a revised P(doom) by the market's close, resolves NO

In ambiguous cases, I'll use my judgment. For example, if Yud were to say something like

idk, not sure I want to make a public P(doom) prediction right now, but if I did, it would be in the neighborhood of 75%. no, wait, maybe 85%. I feel very conflicted about it.

I would treat that as a revision to 0.80.

Get
Ṁ1,000
to start trading!