Cultural Alignment

Exploring real AI risks through the lens of pop culture

Still from Parks and Recreation illustrating T-Dazzle

T-Dazzle

01Scenario

Leslie and Tom rebrand fluoride as “T-Dazzle,” turning an unpopular public-health policy into a 72-percent polling winner without changing the substance.

02AI safety Analogy

An interface optimized around biases can manufacture apparent preference without producing better-informed consent.

03Where the analogy breaks

The tactic supports a beneficial policy, which makes the manipulation-versus-outcome tradeoff especially ambiguous.

AI Risk families

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

AI safety concepts

This scenario is related to the following AI safety concepts.
  1. Value Alignment
  2. AI Disinformation
  3. Sycophancy

Source continuation

More from Parks and Recreation

View all 2 scenarios
  1. Still from Parks and Recreation illustrating Gryzzlbox Knows Too Much

    Parks and Recreation2015

    Gryzzlbox Knows Too Much

Taxonomy neighbors

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

  1. Still from Mad Men illustrating It’s Toasted

    Mad Men2007

    It’s Toasted

    Risk
    Malicious use · Systemic
    Concept
    Value Alignment · AI Disinformation
  2. Still from Pluribus illustrating Pluribus Hand Grenade

    Pluribus2025

    Pluribus Hand Grenade

    Risk
    Systemic
    Concept
    Value Alignment · Sycophancy
  3. Still from Pluribus illustrating The Truthful Charm Offensive

    Pluribus2025

    The Truthful Charm Offensive

    Risk
    Systemic
    Concept
    Value Alignment · Sycophancy