LessWrong AI
· Communities
A Solution to Cryptographic Boxes for Unfriendly AI
SummaryIn 2010, Paul Christiano wrote Cryptographic Boxes for Unfriendly AI, in which he asks how we can sandbox arbitrarily dangerous AIs and recommends Homomorphic Encryption as a potential solution. However, Homomorphic Encryption relies on computational assumptions (it does not provide perfect secrecy) and is extre