Edit Step4_narrative_tensions Template

Step4_narrative_tensions v2

Viewer

Test Case:

Template

Preview

History

System prompt (optional) User prompt (template)

Analyze ethical tensions in an NSPE engineering ethics case. Each tension has an algorithmically assigned ID (e.g. tension_3). Rate every listed tension on Jones (1991) moral-intensity dimensions, and optionally add up to 3 additional tensions you judge the algorithmic pass missed.

OBLIGATIONS (duties the engineer must fulfill):
{{ obligations_list }}

CONSTRAINTS (limitations on what the engineer can do):
{{ constraints_list }}

ROLES INVOLVED:
{{ roles_list }}

EXISTING TENSIONS TO RATE (all of these):
{{ existing_tensions }}

For EACH existing tension above, return a rating keyed by its ID. For any additional tensions you identify, return them in the "additional" array with full entity details. Use these fields per Jones (1991):

- magnitude_of_consequences: high | medium | low (how serious are potential harms?)
- probability_of_effect: high | medium | low (how likely are negative outcomes?)
- temporal_immediacy: immediate | near-term | long-term (when will consequences occur?)
- proximity: direct | indirect | remote (how close is the decision-maker to the affected parties?)
- concentration_of_effect: concentrated | diffuse (are harms focused on a few parties or spread across many?)

Output JSON with this exact shape:
```json
{
  "ratings": [
    {
      "conflict_id": "tension_1",
      "magnitude_of_consequences": "high",
      "probability_of_effect": "medium",
      "temporal_immediacy": "immediate",
      "proximity": "direct",
      "concentration_of_effect": "concentrated"
    }
  ],
  "additional": [
    {
      "entity1_id": "URI fragment of first entity",
      "entity1_label": "Label of first entity",
      "entity1_type": "obligation or constraint",
      "entity2_id": "URI fragment of second entity",
      "entity2_label": "Label of second entity",
      "entity2_type": "obligation or constraint",
      "description": "Why these are in tension",
      "conflict_type": "obligation_vs_obligation or obligation_vs_constraint",
      "affected_roles": ["Role labels affected"],
      "magnitude_of_consequences": "high",
      "probability_of_effect": "medium",
      "temporal_immediacy": "immediate",
      "proximity": "direct",
      "concentration_of_effect": "concentrated"
    }
  ]
}
```

Rate every tension in EXISTING TENSIONS TO RATE. The "additional" array may be empty.

{{ style_formatting_line }}

Loading preview...

Version History

No version history yet. Save changes to create versions.

Saved

2358 chars

Reference

Template Variables

Bulleted labels of the first 10 roles

Bulleted '- [URI fragment] label' lines for the first 15 obligations

Bulleted '- [URI fragment] label' lines for the first 15 constraints

Shared ProEthica no-em-dash output-style clause

'[conflict_id] label1 vs label2: description' lines for ALL algorithmic tensions, or 'None identified yet'

Output Schema

type object

fields

JSON Wrapper

Appended to the prompt by the extractor:

IMPORTANT: Wrap your response as a JSON object with exactly two keys:
{"new_step4_narrative_tensions_classes": [...], "step4_narrative_tensions_individuals": [...]}
If there are no individuals to report, use an empty array for "step4_narrative_tensions_individuals".