Cultural Alignment

Exploring real AI risks through the lens of pop culture

Still from Hunter × Hunter illustrating Nanika’s Hidden Admin CommandScene still / Hunter × Hunter

Nanika’s Hidden Admin Command

Nanika’s public wish interface carries dangerous follow-on costs, but Killua possesses a private command privilege that bypasses the apparent price structure and uses it to restore Gon.

The Zoldycks are containing a poorly understood super-capability while one insider has undocumented administrator access to it.

Nanika’s magical, relationship-dependent rules remain intentionally ambiguous.

AI Risk families

This scenario is an example of the following types of AI risk.
  1. Accidents
  2. Misalignment
  3. Security

AI safety concepts

This scenario is related to the following AI safety concepts.
  1. Capability Evals
  2. AI Control
  3. Power Concentration
  4. Specification Gaming