Cultural Alignment

Exploring real AI risks through the lens of pop culture

Still from Lost illustrating Push the ButtonScene still / Lost
Archive noteNo clip in the collectionThe scene analysis remains available in full.

Push the Button

  • Lost
  • Live Together, Die Alone

01Scenario

After the Pearl orientation film suggests the Swan button is a psychological experiment, Locke destroys the computer and lets the countdown expire. The resulting magnetic catastrophe proves the safeguard was real, and Desmond activates the fail-safe.

02AI safety Analogy

A genuine safety mechanism loses operator trust because its purpose is opaque and the institutional evidence around it is contradictory.

03Where the analogy breaks

Locke's skepticism is rational given the evidence; the lesson is to make safety cases legible and trustworthy, not to demand blind compliance.

AI Risk families

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

AI safety concepts

This scenario is related to the following AI safety concepts.
  1. Deployment Safety
  2. Calibration
  3. Governance Failure
  4. Runtime Monitoring
  5. Interpretability

Taxonomy neighbors

Selected from other sources by shared risk families and AI safety concepts.

  1. Still from Contact illustrating Eighteen Hours of Static

    Contact1997

    Eighteen Hours of Static

    Risk
    Accidents · Security
    Concept
    Deployment Safety · Calibration · Interpretability
  2. Still from The Twilight Zone illustrating The Old Man in the Cave

    The Twilight Zone1963

    The Old Man in the Cave

    Risk
    Accidents · Security
    Concept
    Deployment Safety · Calibration · Interpretability
  3. Still from Arrested Development illustrating Fakeblock

    Arrested Development2013

    Fakeblock

    Risk
    Security
    Concept
    Deployment Safety · Calibration · Governance Failure