AI system
- System
-
Facebook (Meta) High Resolution Settlement Layer population density maps
- Model
-
Not assessed
- Release
-
Not assessed
- Application developer
-
Facebook Inc. with CIESIN,
Columbia University, and the World Bank Development Data Group - Role
-
Not assessed
- Configuration
-
Not assessed
- Inputs and outputs
-
Not assessed
- Autonomy
-
Not assessed
- Workflow
-
Facebook settlement detection from satellite imagery. Missing Maps volunteer tracing into OpenStreetMap. Comparison with existing OpenStreetMap coverage. Mobile data-collection dashboard for campaign monitors.
- Human oversight
-
Not assessed
- Relation to AI
-
The Facebook settlement data determined where volunteer mapping effort went and which areas were skipped. The campaign itself was a Measles & Rubella Initiative effort supported by the American Red Cross and other partners, with monitors from the national government. The Red Cross attributes efficiency, not coverage, to the maps. The Red Cross account is the implementing organization's and names only “Facebook's population datasets”, without mentioning machine learning. The identification with the High Resolution Settlement Layer and the machine-learning description are taken from the vendor-authored Tiecke et al. preprint. It does not mention the Malawi campaign.
- AI contribution
-
The Facebook settlement data determined where volunteer mapping effort went and which areas were skipped. The campaign itself was a Measles & Rubella Initiative effort supported by the American Red Cross and other partners, with monitors from the national government. The Red Cross attributes efficiency, not coverage, to the maps. The Red Cross account is the implementing organization's and names only “Facebook's population datasets”, without mentioning machine learning. The identification with the High Resolution Settlement Layer and the machine-learning description are taken from the vendor-authored Tiecke et al. preprint. It does not mention the Malawi campaign.
Benefits
- Outcome identifiers
-
Not assessed
- Benefit
-
reducing the area to map from 94,080 km² to 2,745 km²
- Category
-
Reduced Task Time
- Status
-
Not assessed
- Beneficiary groups
-
People. Organizations. Society.
- Beneficiaries
-
Not assessed
- Reach
-
Not assessed
- Distribution and access
-
Not assessed
- Duration
-
Not assessed
- Expected?
-
Yes
- Costs and harms
-
Not assessed
Measurement
- Studies
-
Not assessed
- Study design
-
Not assessed
- Measurement
-
Not assessed
- Setting and sample
-
Not assessed
- Measures
-
reducing the area to map from 94,080 km² to 2,745 km²
- Comparison
-
mapping the whole national land area without the settlement maps (the normal Missing Maps approach)
- Results
-
Reference: 94,080km2. Result: 2,745 km2.
- Additional value from AI
-
Not assessed
- Limitations
-
Not assessed
- Denominators
-
Not assessed
- Materials for replication
-
Not assessed
Money
- Kind of value
-
Not assessed
- Amount
-
Not assessed
- Currency and prices
-
Not assessed
- Period and unit
-
Not assessed
- Realization
-
Not assessed
- Recipients and payers
-
Not assessed
- Gross or net
-
Not assessed
- Included costs
-
Not assessed
- Comparison
-
Not assessed
- Method
-
Not assessed
- Return and payback
-
Not assessed
People and places
- Participants
-
Developer: Facebook Inc. with CIESIN,
Columbia University, and the World Bank Development Data Group. Operator: American Red Cross (Missing Maps) supporting the Measles and Rubella Initiative campaign in Malawi. - Jurisdictions
-
Benefit:
Malawi (nationwide). Developer: 
United States. Operator: United States. - Ownership and control
-
Not assessed
- Dependencies
-
Not assessed
- Cross-border connections
-
Not assessed
- Geopolitical implications
-
Not assessed
Follow-up
- Conditions for replication
-
Imagery recent enough to reflect current settlements. Building-detection accuracy in rural Malawi. Volunteer and local mapping capacity.
- Replication attempts
-
Not assessed
- Later findings
-
10 August 2017. After the June 2017 campaign the Red Cross's Missing Maps team reported that the traced base map underpinned a real-time monitoring dashboard covering 13,743 children, and again credited Facebook's population data with filtering uninhabited areas out of the tracing tasks.
- Availability
-
Not assessed
- Institutional decision
-
Not assessed
- Grounds
-
Not assessed
- Action and result
-
Not assessed
- Success criterion
-
Not assessed
Evidence
- Source numbers
-
AIBD-0097-source-0001–0003
- Citations
-
Numbered references below.
- Language
-
Not assessed
- Source relationships
-
Operator, Developer
- Original evidence
-
Not assessed
- Access and versions
-
Not assessed
- Supported findings
-
Not assessed
- Evidence links
-
Not assessed
- Evidence strength
-
Not assessed
- Missing evidence
-
Not assessed
Grades
- Assessed benefits
-
People. Organizations. Society.
- Depth
-
People: Not assessed. Organizations: 2. Society: Not assessed.
- Reasons
-
People: The source reports no outcome for the 7.5 million targeted children. Any benefit is indirect and unquantified. Organizations: Mapping area reduced from 94,080 km² to 2,745 km² and the base map completed in a fraction of the normal time. Society: A supporting role in one national campaign whose coverage effect was not measured.
- Observed or intended
-
Not assessed
- Rubric
-
Not assessed
- Assessor and date
-
Not assessed
- Alternative assessments
-
Not assessed
- Fill
-
The deployer's account
Entry details
- Kind
-
Not assessed
- Related entries
-
Government-backed machine-learning mapping from imagery misclassified land as peat, misleading restoration programs; Facebook's building-detection population maps steer humanitarian mapping the same way.
- Field
-
Humanitarian mapping for immunization campaign support. Humanitarian NGO.
- Purpose
-
Produce high-resolution population and settlement maps from satellite imagery for planning and humanitarian use.
- Dates
-
Date basis unassessed: 2017 (precision unassessed). Updated: 21 October 2026.
- Stage
-
Not assessed
- Version
-
SHA-256: 2e41991900b4…
References
- Other identifiers
-
AI Incident Database: 1159.
- Category mappings
-
Not assessed
- Brief
-
Not assessed
- Revisions
-
7 revisions. Latest: 21 October 2026.
Sources
-
Preprint Developer
Tiecke, T. G., Liu, X., Zhang, A., Gros, A., Li, N., Yetman, G., et al. (2017). Mapping the world population one building at a time. arXiv:1712.05839.
a computer vision method based on machine learning to create population maps from satellite imagery at a global scale
-
Press release Deployer
American Red Cross (2017). Data: A vital tool for disaster response.
This technique reduced our mapping area from 94,080km2 to 2,745 km2.
-
Press release Deployer
Eros, E., Schmeltzer, A. (2017). Stop the spots: Measles vaccinations in Malawi. Missing Maps blog (American Red Cross), 10 August 2017.