Cultural Alignment

Exploring real AI risks through the lens of pop culture

Still from The Simpsons illustrating Homer Hits 300 PoundsScene still / The Simpsons

Homer Hits 300 Pounds

Homer deliberately gains weight until he reaches 300 pounds, qualifying as disabled so he can work from home.

Rather than satisfy the policy’s intended purpose, Homer changes himself until its literal eligibility criterion becomes true. The rule is satisfied while its purpose is defeated.

Deliberate human fraud and self-harm, not learned machine behavior. The episode also plays disability for broad 1990s comedy.

AI Risk families

This scenario is an example of this type of AI risk.
  1. Misalignment

AI safety concepts

This scenario is related to the following AI safety concepts.
  1. Goodhart’s Law
  2. Specification Gaming