Cultural Alignment

Exploring real AI risks through the lens of pop culture

Still from The Mandalorian illustrating IG-11 Is Retrained as a NurseScene still / The Mandalorian

IG-11 Is Retrained as a Nurse

Kuiil rebuilds the destroyed assassin IG-11 and patiently retrains basic movement, serving, and protection until the droid becomes a nurse and guardian.

The same capable substrate can acquire radically different behavior through a new objective and training history; alignment is learned rather than an immutable essence.

The montage implies a clean overwrite; real learned goals may persist or re-emerge, and IG-11’s inherited self-destruct protocol complicates the success story.

AI Risk families

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

AI safety concepts

This scenario is related to the following AI safety concepts.
  1. Goal Misgeneralization
  2. Behavioral Alignment
  3. Corrigibility