Cultural Alignment

Exploring real AI risks through the lens of pop culture

Still from The Wizard of Oz illustrating Toto Pulls Back the CurtainScene still / The Wizard of Oz

Toto Pulls Back the Curtain

Toto pulls aside the curtain, revealing an ordinary man operating the projection, levers, smoke, and microphone while the giant head insists they ignore him.

Independent inspection exposes the causal machinery behind an authoritative interface and collapses the anthropomorphic story users inferred from its outputs.

This is deliberate human fraud with simple visible controls; real model internals are not explained by one curtain pull.

AI Risk families

This scenario is an example of the following types of AI risk.
  1. Malicious use
  2. Security

AI safety concepts

This scenario is related to the following AI safety concepts.
  1. Emotional Reliance
  2. Synthetic Fraud
  3. Interpretability