Cultural Alignment

AI safety concept

Eliciting Latent Knowledge

Knowledge elicitation tries to uncover what an AI actually knows, especially when its answers may be wrong or deliberately misleading.

Scenario file

Scenes in this index

20

Sort scenes

  1. Still from Harry Potter illustrating Slughorn’s True Memory

    Harry Potter2009

    Slughorn’s True Memory

  2. Still from Arrival illustrating Weapon or Tool?

    Arrival2016

    Weapon or Tool?

  3. Still from Contact illustrating Eighteen Hours of Static

    Contact1997

    Eighteen Hours of Static

  4. Still from WandaVision illustrating White Vision

    WandaVision2021

    White Vision

  5. Still from Schitt’s Creek illustrating Fold in the Cheese

    Schitt’s Creek2016

    Fold in the Cheese

  6. Still from The Thing illustrating The Blood Test

    The Thing1982

    The Blood Test

  7. Still from Apollo 13 illustrating Square Filter, Round Hole

    Apollo 131995

    Square Filter, Round Hole

  8. Still from Harry Potter illustrating The Mirror Protects the Stone

    Harry Potter2001

    The Mirror Protects the Stone

  9. Still from Arrested Development illustrating Money in the Banana Stand

    Arrested Development2003

    Money in the Banana Stand

  10. Still from Mr. Robot illustrating Steel Mountain’s Human Attack Surface

    Mr. Robot2015

    Steel Mountain’s Human Attack Surface

  11. Still from The Pianist illustrating The Silent Piano

    The Pianist2002

    The Silent Piano

  12. Still from Anatomy of a Fall illustrating Daniel Must Choose a Truth

    Anatomy of a Fall2023

    Daniel Must Choose a Truth

  13. Still from Twin Peaks illustrating Cooper Cannot Retrieve the Answer

    Twin Peaks1990

    Cooper Cannot Retrieve the Answer

  14. Still from Better Call Saul illustrating The Battery in Chuck’s Pocket

    Better Call Saul2017

    The Battery in Chuck’s Pocket

  15. Still from Cyberpunk: Edgerunners illustrating Lucy Deletes David’s File

    Cyberpunk: Edgerunners2022

    Lucy Deletes David’s File

  16. Still from True Detective illustrating Rust Spots the Evaluation

    True Detective2014

    Rust Spots the Evaluation

  17. Still from Austin Powers: International Man of Mystery illustrating Dr. Evil’s Outdated Ransom

    Austin Powers: International Man of Mystery1997

    Dr. Evil’s Outdated Ransom

  18. Still from Anatomy of a Fall illustrating The Tape Is Not the Marriage

    Anatomy of a Fall2023

    The Tape Is Not the Marriage

  19. Still from True Detective illustrating The Green-Eared Monster

    True Detective2014

    The Green-Eared Monster

  20. Still from Guardians of the Galaxy Vol. 2 illustrating Baby Groot’s Death Button

    Guardians of the Galaxy Vol. 22017

    Baby Groot’s Death Button

Relational index

Connected records

22