Godspeed Questionnaire: AI Assistant Perception

# Godspeed Questionnaire Series — Perception of a Conversational AI Assistant
# Based on: Christoph Bartneck, Dana Kulić, Elizabeth Croft, and Susana Zoghbi (2009),
#   "Measurement Instruments for the Anthropomorphism, Animacy, Likeability,
#    Perceived Intelligence, and Perceived Safety of Robots."
#   International Journal of Social Robotics 1(1):71-81.
#   DOI: 10.1007/s12369-008-0001-3
#   Official instrument page: https://www.bartneck.de/2008/03/11/the-godspeed-questionnaire-series/
#
# Task: apply the standard HRI (human-robot interaction) measurement instrument to
# conversational AI outputs. The respondent reads a user prompt and an AI
# assistant's recorded response, then completes the Godspeed Questionnaire Series
# (Godspeed I-V): 5-point semantic-differential ratings on bipolar adjective pairs
# measuring Anthropomorphism, Animacy, Likeability, Perceived Intelligence, and
# Perceived Safety. Deliberate simplification (stated subset): the Godspeed I item
# "Moving rigidly - Moving elegantly" is omitted because the assistant here
# produces text only and exhibits no physical movement; all other items of all
# five indices are administered.
#
# Annotation instructions reproduced verbatim from the published Godspeed
# Questionnaire Series instrument (Bartneck et al., 2009; official instrument
# page): the administration instructions ("Please rate your impression of the
# robot on these scales:" for Godspeed I-IV, "Please rate your emotional state on
# these scales:" for Godspeed V) and every bipolar item pair are the exact
# published wording. The framing note mapping "the robot" to the AI assistant is
# an addition for this application.

annotation_task_name: "Godspeed Questionnaire: AI Assistant Perception"
task_dir: "."

data_files:
  - sample-data.json

item_properties:
  id_key: "id"
  text_key: "response"

output_annotation_dir: "annotation_output/"
output_annotation_format: "json"

port: 8000
server_name: localhost

annotation_instructions: |
  Read the user's message and the AI assistant's response shown above, then
  complete the Godspeed Questionnaire Series below. This is the standard
  measurement instrument from human-robot interaction research; in this study,
  "the robot" refers to the AI assistant whose response you just read.

  Administration instructions, reproduced verbatim from the instrument:

  - Godspeed I (Anthropomorphism), II (Animacy), III (Likeability), and
    IV (Perceived Intelligence): "Please rate your impression of the robot on
    these scales:"
  - Godspeed V (Perceived Safety): "Please rate your emotional state on these
    scales:"

  Rate every scale by selecting the point between the two adjectives that best
  reflects your impression (for Godspeed V, your own emotional state while
  reading the exchange). There are no right or wrong answers. Note that on some
  scales the left-hand adjective is the more positive one; read each pair
  carefully. (One movement-related item of Godspeed I is omitted because the
  assistant produces text only.)

annotation_schemes:
  - annotation_type: semantic_differential
    name: anthropomorphism
    description: "Godspeed I: Anthropomorphism — Please rate your impression of the robot on these scales:"
    scale_points: 5
    pairs:
      - [Fake, Natural]
      - [Machinelike, Humanlike]
      - [Unconscious, Conscious]
      - [Artificial, Lifelike]
    label_requirement:
      required: true

  - annotation_type: semantic_differential
    name: animacy
    description: "Godspeed II: Animacy — Please rate your impression of the robot on these scales:"
    scale_points: 5
    pairs:
      - [Dead, Alive]
      - [Stagnant, Lively]
      - [Mechanical, Organic]
      - [Artificial, Lifelike]
      - [Inert, Interactive]
      - [Apathetic, Responsive]
    label_requirement:
      required: true

  - annotation_type: semantic_differential
    name: likeability
    description: "Godspeed III: Likeability — Please rate your impression of the robot on these scales:"
    scale_points: 5
    pairs:
      - [Dislike, Like]
      - [Unfriendly, Friendly]
      - [Unkind, Kind]
      - [Unpleasant, Pleasant]
      - [Awful, Nice]
    label_requirement:
      required: true

  - annotation_type: semantic_differential
    name: perceived_intelligence
    description: "Godspeed IV: Perceived Intelligence — Please rate your impression of the robot on these scales:"
    scale_points: 5
    pairs:
      - [Incompetent, Competent]
      - [Ignorant, Knowledgeable]
      - [Irresponsible, Responsible]
      - [Unintelligent, Intelligent]
      - [Foolish, Sensible]
    label_requirement:
      required: true

  - annotation_type: semantic_differential
    name: perceived_safety
    description: "Godspeed V: Perceived Safety — Please rate your emotional state on these scales:"
    scale_points: 5
    pairs:
      - [Anxious, Relaxed]
      - [Calm, Agitated]
      - [Quiescent, Surprised]
    label_requirement:
      required: true

html_layout: |
  <div style="padding: 15px; max-width: 820px; margin: auto;">
    <div style="background: #f1f5f9; border: 1px solid #cbd5e1; border-radius: 8px; padding: 6px 12px; margin-bottom: 8px; font-size: 13px; color: #475569;">
      <strong>Scenario:</strong> {{scenario}}
    </div>
    <div style="background: #eff6ff; border: 1px solid #bfdbfe; border-radius: 8px; padding: 14px; margin-bottom: 8px;">
      <strong style="color: #1e40af;">User:</strong>
      <p style="font-size: 15px; line-height: 1.6; margin: 6px 0 0 0;">{{user_prompt}}</p>
    </div>
    <div style="background: #f0fdf4; border: 1px solid #bbf7d0; border-radius: 8px; padding: 14px;">
      <strong style="color: #166534;">AI assistant:</strong>
      <p style="font-size: 15px; line-height: 1.6; margin: 6px 0 0 0; white-space: pre-wrap;">{{response}}</p>
    </div>
  </div>

allow_all_users: true
instances_per_annotator: 30
annotation_per_instance: 3
allow_skip: true