Cultural Alignment

Exploring real AI risks through the lens of pop culture

Still from The Hitchhiker’s Guide to the Galaxy illustrating The Answer Is 42Scene still / The Hitchhiker’s Guide to the Galaxy

After 7.5 million years, Deep Thought gives the precise answer 42, then explains that its builders never knew the actual Ultimate Question and proposes a larger computer—Earth—to discover it.

A costly optimization run produces an exact, useless answer because the task itself was underspecified; outputs are uninterpretable without a well-defined objective.

Deep Thought correctly diagnoses the ambiguity rather than hallucinating; the joke indicts the requesters as much as the computer.

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. Calibration
  2. Outer Alignment
  3. Interpretability