Target Audience: Senior technology professionals, policy makers, program managers designing digital governance systems
Prerequisites: Workshop 1 or equivalent knowledge of digital ethics frameworks
Materials Needed: Policy documents for analysis, laptops for governance framework design, templates for consent processes
By the end of this workshop, participants will be able to:
ASHA Data Collection: Community health workers use tablets to collect sensitive health data, including HIV status, domestic violence cases, and mental health information. Data flows to district officials, researchers, and insurance companies.
School Feeding Program: Biometric systems track children's meal consumption for program monitoring. Data includes attendance patterns, nutritional status, and family socioeconomic indicators stored centrally.
Migrant Worker Registration: Digital platform registers interstate migrants, collecting Aadhaar, employment history, and location tracking for COVID contact tracing and labor law compliance.
Development Privacy Challenges: Privacy in development programming operates in contexts of vulnerability, power imbalances, and limited alternatives. Traditional consent models often fail when people have no choice but to participate in programs for survival.
Core Principles:
Challenge: Design a consent process for the ASHA data collection scenario above.
Context Analysis:
Design Challenge: Create a consent framework that addresses:
Output: 3-minute presentation of your consent design with justification for key choices.
What starts as program monitoring can evolve into comprehensive surveillance systems. Development programs create unprecedented data about vulnerable populations, which can be repurposed for social control, political monitoring, or commercial exploitation.
Surveillance Spectrum: Not all monitoring is surveillance, but the line is often blurry. Understanding this spectrum helps identify when development programs cross ethical boundaries.
| Purpose | Legitimate Monitoring | Surveillance Risk | Mitigation Strategies |
|---|---|---|---|
| Program Accountability | Tracking benefit delivery and outcomes | Creating individual behavioral profiles | Aggregate data, limited retention |
| Service Delivery | Ensuring services reach intended beneficiaries | Monitoring daily activities and movements | Purpose limitation, data minimization |
| Research & Learning | Understanding program effectiveness | Long-term tracking without consent renewal | Dynamic consent, anonymization |
| Fraud Prevention | Detecting duplicate or false beneficiaries | Behavioral scoring and social network analysis | Algorithmic transparency, human review |
Initial Program (2019): A state health department launches digital nutrition tracking for pregnant women. ASHA workers use app to record weight gain, clinic visits, and dietary counseling.
Data Scope Expansion (2020): System expanded to include family composition, income sources, migration patterns, and social network connections for "better targeting."
Secondary Use (2021): Police access system during communal tensions to identify families with "irregular" movement patterns. Social welfare department flags families with non-compliance for "intensive counseling."
Commercial Use (2022): Insurance companies request access to health behavior data for risk assessment. Microfinance institutions want to use compliance scores for loan decisions.
Resistance and Pushback (2023): Women's groups report harassment based on system data. ASHA workers complain of becoming "surveillance agents." Some communities begin avoiding the program.
Ethical Failures:
Question: Who controls data collection, storage, and use decisions?
Analysis: Examine asymmetries between data subjects and data controllers
Question: Who designs algorithms and sets parameters for automated decisions?
Analysis: Assess transparency and contestability of algorithmic systems
Question: Who controls the digital infrastructure and sets the rules?
Analysis: Evaluate dependency and lock-in effects
Question: How does data create or redistribute economic value?
Analysis: Track who benefits financially from data extraction
Scenario: A state government proposes a "Smart Safety Net" system that integrates data from PDS, MGNREGA, health programs, and educational services to create comprehensive household profiles for better targeting.
System Features:
Assessment Questions (work in pairs):
Power Analysis:
Risk Assessment:
Governance Questions:
Output: Risk matrix with high/medium/low ratings for different surveillance concerns and specific mitigation recommendations.
Platform Reality: Increasingly, development programs rely on digital platforms, from WhatsApp groups for farmer extension to proprietary software for benefit delivery. These platforms shape how development happens and who has voice in development processes.
Key Governance Dimensions:
| Governance Area | Key Questions | Development Implications |
|---|---|---|
| Access & Inclusion | Who can join? What are the barriers? | Digital divides exclude marginalized groups |
| Content Moderation | What speech is allowed? Who decides? | Critical feedback may be suppressed |
| Data Governance | How is user data collected and used? | Program participants become data products |
| Algorithmic Curation | How does the algorithm decide what you see? | Information asymmetries affect decision-making |
| Economic Model | How does the platform make money? | User attention and data become commodified |
| Exit Rights | Can users leave? What happens to their data? | Platform lock-in creates dependency |
Hypothetical case (fictional company and app, hypothetical figures). Charchamitra and its owner, Charchamitra Networks, are invented for this exercise.
Context: Agricultural extension services in one state use groups on Charchamitra, a commercial messaging app, to connect 10,000+ farmers with experts, weather information, and market prices.
Initial Success: Farmers report 40% improvement in access to timely agricultural advice. Extension officers can reach more farmers efficiently.
Governance Challenges Emerged:
Platform Governance Failures:
Challenge: Design governance mechanisms for a new digital platform connecting urban youth with rural livelihood opportunities.
Platform Description:
Governance Design Task: Create mechanisms for:
1. Democratic Participation (3 minutes):
2. Accountability Mechanisms (3 minutes):
3. Value Distribution (3 minutes):
4. Exit Rights (3 minutes):
Output: One-page platform governance charter with specific mechanisms for each governance area.
AI Proliferation: From predictive models for malnutrition risk to chatbots providing agricultural advice, AI systems are increasingly embedded in development programs. These systems require specific governance frameworks beyond general algorithmic accountability.
Impact Assessment: Systematic evaluation of AI system effects on different groups
Community Consultation: Meaningful engagement with affected communities in design
Bias Testing: Technical audits for discriminatory outcomes
Human Oversight: Qualified humans can review and override AI decisions
Explainability: AI decisions can be explained in understandable terms
Gradual Rollout: Phased deployment with continuous monitoring
Continuous Monitoring: Ongoing assessment of real-world performance
Appeal Mechanisms: Processes for challenging AI decisions
Model Updates: Regular retraining and bias correction
Documentation: Complete records of AI system development and deployment
Stakeholder Engagement: Ongoing dialogue with affected communities
Retirement Planning: Clear processes for ending AI system use
Initiative: A state health department launches an AI chatbot to provide 24/7 maternal health advice to pregnant women in rural areas, available in the state language and English.
AI System Features:
Initial Metrics (6 months): 50,000+ users, 89% satisfaction rate, 23% reduction in emergency visits
Problems Discovered (12 months):
Crisis Point: Two preventable deaths linked to incorrect AI advice led to public backlash and program suspension
Governance Failures:
Scenario: You're designing governance for an AI system that predicts which children are at highest risk of dropping out of school, to prioritize intervention resources.
AI System Details:
Governance Planning Task (work in groups of 3-4):
Impact Assessment (4 minutes):
Accountability Mechanisms (4 minutes):
Community Engagement (4 minutes):
Implementation Safeguards (3 minutes):
Output: AI governance checklist with specific procedures for each stage of system lifecycle.
Before Implementation:
During Implementation:
Ongoing Governance:
Indian Policy Frameworks:
Technical Implementation Tools:
Organizational Capacity Building:
Next Steps in ImpactMojo:
This handout is part of the ImpactMojo 101 Knowledge Series
Licensed under CC BY-NC-ND 4.0 • Free to use with attribution • www.impactmojo.in
For privacy impact assessment templates, AI governance checklists, and platform governance frameworks, visit the ImpactMojo platform.
