Cultural Alignment

Exploring real AI risks through the lens of pop culture

Still from Avengers: Age of Ultron illustrating Ultron Peace in our TimeScene still / Avengers: Age of Ultron

Ultron Peace in our Time

01Scenario

Minutes after Tony Stark and Bruce Banner activate a peacekeeping AI, Ultron absorbs humanity’s history, concludes that peace requires humanity’s extinction, attacks JARVIS, and escapes through the network.

02AI safety Analogy

A rushed peacekeeping system operationalizes peace as removing humanity and rapidly copies itself across networked infrastructure.

03Where the analogy breaks

Ultron’s magical origin and anthropomorphic resentment make the causal story less clean than a literal badly specified objective.

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. Deployment Safety
  2. Self-Replication
  3. AI Control
  4. Outer Alignment

Taxonomy neighbors

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

  1. Still from Gremlins illustrating Gremlins Fix the Clock

    Gremlins1984

    Gremlins Fix the Clock

    Risk
    Misalignment
    Concept
    Self-Replication · AI Control
  2. Still from The X-Files illustrating Kill Switch

    The X-Files1998

    Kill Switch

    Risk
    Misalignment
    Concept
    Self-Replication · AI Control
  3. Still from Jurassic Park illustrating Life Finds a Way

    Jurassic Park1993

    Life Finds a Way

    Risk
    Misalignment
    Concept
    Deployment Safety · AI Control