AVPN AI for Good Fellowship Resources
The following resources are available on the data.org platform for Fellows who wish to explore topics more deeply between sessions – or recommend resources to their colleagues. All resources are optional, and this page will be updated after each session. Each session is designed to stand alone for Fellows who have not completed the additional reading. See the full AVPN AI for Good Fellowship syllabus here.
ARTPARK Invests in People and Process to Improve Health Outcomes data.org
Background on this session’s guest organisation. Covers ARTPARK’s dengue risk-prediction platform, now embedded in public health infrastructure across Karnataka, Pune, and Pimpri-Chinchwad — impacting 80M+ people — and the equity problem built into any prediction model: areas without robust reporting tend to get missed by the very systems meant to protect them.
UNESCO’s Recommendation on the Ethics of Artificial Intelligence: Key Facts UNESCO
The source behind this session’s “meaningful human oversight” reference. Includes a short worked example — an AI system auto-declining a loan application with no human reviewer and no appeals process — that’s effectively Rung Four from tonight’s Delegation Ladder.
Myna Mahila Foundation data.org
Another AI use case for your review. A text-based AI chatbot delivering sexual and reproductive health information in Hindi, Hinglish, and Marathi, aimed at dispelling misconceptions among women with limited digital access.
Should I Be Using AI for This? Tech Matters / Jim Fruchterman
The source of the Nonprofit AI Treasure Map used in this session, with the full reasoning behind each gate.
Digital Green: Farmer.Chat Digital Green / OpenAI
Another AI use case for your review. The organisation’s own account of Farmer.Chat, including the decision to deploy through extension agents rather than direct to farmers. Vendor-published—read alongside the independent evaluation below.
Video-based support for small-scale farmers J-PAL
What the video model achieved before AI, which components were deliberately preserved, and how Farmer.Chat is being independently evaluated with IFPRI.
How to respectfully use and inform communities about PII collected about them data.org
Directly relevant to Scenario B; covers consent, transparency, and community communication about personal data.
How to improve staff data literacy data.org
Addresses the People dimension of the DMA.
Ethical AI in Practice data.org
A practical course on fundamental concepts of AI, how AI works, and how AI makes a difference in government and social development work.
Adopting Responsible Data Management data.org
Covers responsible data practices across a program’s lifecycle, with a social impact focus.
AI Ethics Guidelines UNESCO
The UNESCO Recommendation on the Ethics of AI — the first global normative framework.
DMA in Action: Generate Health data.org
Illustrates how a social impact organisation used the DMA to benchmark readiness, identify gaps, and prioritise investments in data capability.
How to prioritise efforts when strengthening data maturity data.org
A practical guide for translating DMA scores into prioritised action.
Should your social impact organization use AI? IDinsight / Sid Ravinutala & Marc Shotland
A candid guide for social sector leaders navigating pressure to adopt AI without a clear strategy.
How AI Evaluation Works in Practice: Insights from Implementers IDinsight / Isha Fuletra & Suzin You
Draws on interviews with practitioners deploying AI across health, social protection, justice, and behaviour change to show how evaluation actually happens under real constraints.
AI Evaluation in the Social Sector: A Living Playbook The Agency Fund
An open, evolving playbook for practitioners working with generative AI in the social sector, built on the four-level evaluation framework—model, product, user, and impact—with worked examples from eight GenAI projects in health, education, and agriculture and decision guides.
