Will Superalignment succeed, according to a majority of Eliezer, Nate, Paul, John, David.
Mini
7
Ṁ1248
2028
4%
chance

Superalignment is an agenda at OpenAI attempting to "solve the core technical challenges of superintelligence alignment in four years".
Daniel Filan's AXRP podcast interview with Jan Leike on superalignment

If at least three of the judges believe that the superalignment agenda has been successful in "solv[ing] the core technical challenges of superintelligence alignment" by the end of 2027, this market will resolve YES.
Otherwise it resolves NO.

The judges for this question are:
Eliezer Yudkowsky
Nate Soares
Paul Christiano
John Wentworth
David Krueger

Notes:
The question refers to the agenda, not the team. What that means is up to the individual judges to decide. However, the intent of this question is to predict the success/failure of the agenda - not whether it happens to succeed at OpenAI, or under Jan/Ilya.

The end of year cutoff is to allow ample time for any materials published by the superalignment team to be read and considered by the judges.

For the purposes of this question, a judge believes the agenda succeeded if they say as much publicly (e.g. on the alignment forum, on lesswrong, on twitter). If there's ambiguity, then the judge's direct answer to [should their vote on this question be for yes/no?] always takes precedence.
If they say nothing, that counts as a 'no' vote.

In the unlikely event that any of the judges dies before the market closes, they'll be replaced by someone I guess to have a similar bar for superalignment success. (or if I die too, someone else's best guess!)

Get
Ṁ1,000
and
S1.00