On: Safe Superintelligence Inc.

[ben_w]

That's why it's a problem.

An AI can be anywhere on that axis, and we don't really know what we're doing in order to prevent it being as I have described.