causal
More ways to build misaligned than aligned superintelligence
Probabilistically, there are more ways to build a superintelligent AI that is unaligned than aligned, so if we have not figured out the principle by which we would align it, the idea that we will achieve alignment by chance is far-fetched—and the AI lab principals themselves estimate the probability of destroying everything at 10-30%, when a nuclear reactor's acceptable risk threshold is one in a million per year.
causalpending
Speaker
Sam HarrisEvidence Quote
“there are more ways to build super intelligent ai that that are unaligned than aligned so if we haven't figured out the principle by which we would align it the idea that we're going to do it by chance seems far-fetched”
Created: 6/14/2026, 2:01:37 AM
Your notes
Loading notes...