= 27d567bc2d48d9f40e9ccc20ea370b15)
VAuroch comments on Some alternatives to “Friendly AI” - Less Wrong Discussion
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (44)
Computer security might be a good metaphor, at least in some contexts (particularly institutional contexts). Secure AI or Human-Secure AI gets across the general quality of high stakes with more appropriate connotations than High-Assurance and less impossible.
It also has the bonus that it draws an implicit connection to mathematical security, since insofar as we can say anything about eventual implementation of a Friendly/Safe/Secure AGI, it's fair to say that it's extremely likely to rely on advances descending from current techniques in provably-secure computing. Also the bonus that talking up security is frequently a good way to get Organizations With Money to start reaching for their wallets.
In this context "secure AI" has the connotations of being well-protected against hackers.
Well, that should be a necessary if not sufficient condition. Do you think it would dominate other interpretations to laymen/nonexperts?
I think that "secure" generally has the connotations of being secure against external threats. A "secure facility" is one into which outsiders cannot get into, but it may well contain nuclear bombs.