CrowdAI - Fireline & Spot Fire Identification
Fabric Score 4.2
Civil SocietyPublic ServicesCrowdAINon-GenAI modelProduct Owner ×1
Workflow Diagram
Fabric Score
Task
During an aerial flyover of active fires, identify fire line and spot fires in real-time automatically.
Intent
The AI system improves fireline identification at a larger scale.
AI Workflow
Input
During fire season, aerial flights are conducted during active fires.
Process
The AI system automatically extracts the edges of fireline and spot fires.
Output
Firefighter sees GPS location of fire overlaid on a software product that is on firefighters cell phones.
Human Oversight Level
Human-Approved AI
Institutional Oversight Examples
- •Video footage stays proprietary
- •GPS location is pushed into a user interface owned by the firefighters
Risk
Aerial data collection is necessary and is the bigger limitation to firefighting.
Output Modification Telemetry
- Irrelevant information29%
- Unfaithful outputs (policy, tone, or style misalignment)21%
- Missing or incomplete information21%
- Factual errors / hallucinations13%
- Internal inconsistency11%
- Poor usability (unclear language)5%
Transversal Metrics
Grouped by Fabric dimension.
Efficacy
Accuracy98 %
Efficiency
Modification Rate7 %
Modification Time8 min/output
Verification Time2 min
Rejection Rate3 %
Operational Friction4 hrs/wk
Implementation Overhead500 hrs
Governance Overhead100 hrs
Time to Launch40 wks
Value
Effort Reduction85 %
Risk
Reliability2 incident/month
Autonomy84 %