How AI Will Reshape Global Development Beyond the SDGs

Abstract

The panel explored how artificial intelligence can be leveraged to accelerate the transformation of global development systems that go beyond the Sustainable Development Goals (SDGs). Participants examined AI’s role in disaster‑risk governance, inclusive finance, educational access, agricultural advisory, and public‑sector capacity‑building. The discussion moved from concrete examples of AI‑driven solutions to higher‑level debates about equity, governance, “diffusion pathways,” and the risks of concentration of power. Recommendations centred on open‑source models, public‑sector investment, cross‑regional collaborations, and the need to embed human judgment, linguistic diversity, and democratic oversight in AI deployment.

Detailed Summary

• The session opened with a brief welcome and a reminder that the 2030 deadline for the SDGs is approaching amid intense economic, social, and geopolitical pressures.
• The moderator highlighted the central question: How can AI reshape development pathways in a way that is equitable, scalable, and goes beyond the SDG framework?

---

2. First Panel – AI for Disaster‑Risk Governance & Structural Transformation

2.1 Amandeep Singh Gill – Strengthening Disaster‑Risk Governance

• Emphasised that AI tools must reinforce, not destabilise, existing disaster‑risk governance systems.
• Warned against misinformation and disinformation that could arise from poorly designed AI applications.
• Cited the Sendai Framework: mortality has fallen ~50 % over recent decades, yet housing loss and economic impact are soaring.
• Called for AI to be used for long‑term housing policy and infrastructure resilience, not merely to address visible, short‑term problems.

2.2 Kamal Kishore – Inclusion, Affordability, and Open‑Source Models

• Stated that success in an AI‑driven world must be measured by inclusion, affordability, and scale.
• Noted that ~70 % of AI outputs currently originate from large corporations concentrated in five regions, marginalising startups, SMEs, and emerging economies.
• Described his own legal‑tech AI startup, which uses smaller, open‑source, fine‑tuned models (embedding & RAG architectures) with domain‑specific data to keep tools affordable.
• Reported impact: >1 million students and researchers in India and 100+ universities have benefited.
• Outlined three policy levers for governments:
  1. Subsidised compute for startups,
  2. Support for lightweight/open models,
  3. Investment in high‑quality data sets and AI talent pipelines.
• Stressed two non‑negotiable principles:
  • Human judgment must stay central; AI should augment, not replace, expertise.
  • Upskilling the workforce must be a joint public‑private endeavour.

2.3 Sangbu Kim (World Bank) – “Small AI” for Immediate Impact

• Acknowledged the usual prerequisites for AI (energy, connectivity, data, talent) but argued waiting for perfection delays benefits.
• Introduced the concept of “small AI” – practical, affordable, locally‑relevant solutions that can be deployed despite resource constraints.
• Provided examples of AI‑enabled services for farmers, teachers, students, patients, and doctors.
• Emphasised that developing good practices now is the “good starting point” for wider adoption.

2.4 Shankar Maruwada (EkStep Foundation) – Diffusion Pathways & the 100‑Pathway Goal

• Shared a visionary “dream”: by 2040 a 30‑year‑old woman in rural Nigeria could win a Nobel Prize thanks to AI‑mediated access to global research in her native language.
• Introduced the idea of a “diffusion pathway” – the missing middle between AI invention, innovation, and impact.
• Illustrated rapid rollout timelines:

Project	Deployment Time
Maharashtra agriculture AI advisory	9 months
Bharat Vistar	6 months
Ethiopia Open Agrinet	3 months
Amul (milk cooperative) AI advisory	3 weeks

• Announced a call to create 100 such pathways by 2030 across sectors and continents, with partners like Google, World Bank, UN, etc.

2.5 Vilas Dhar (Patrick J. McGovern Foundation) – Funding, Ecosystem, and Democratic Governance

• Noted that the foundation has > $500 million invested in “AI for Good.”

• Critiqued the “trickle‑down” model of AI diffusion, arguing that concentrating resources in a few pilots does not empower the many.

• Proposed three strategic actions:

  1. Match private‑sector AI investment with public‑sector political will and capital.
  2. Define a rights‑based framework (norms, principles, values) to guide AI development.
  3. Shift agency from technologists/regulators to a broad public conversation, ensuring democratic ownership of AI‑driven development.

• Emphasised that development by 2030 must be co‑created by all stakeholders, not delivered by a privileged few.

---

3. Transition – Photo‑op & Moderator Remarks

• A brief photo‑op with all panelists was organised.
• The moderator thanked participants for staying on time, highlighted the “possibilities and opportunities” discussed, and announced the second part of the town‑hall.

---

4. Second Panel – Inclusion, Risk, and Responsibility

(This segment introduced new speakers who were not on the original speaker list but were part of the recorded session.)

4.1 Bipin Prit Singh – AI‑Enabled Financial Inclusion (MobiQuick)

• Described MobiQuick’s fintech ecosystem: digital payments, loans, investments, advisory – used by >180 million people and 4.5 million small businesses in India.

• Identified three pillars for AI‑driven financial inclusion:

  1. Ease of Use – AI can enable native‑language interfaces, moving beyond English‑only apps.
  2. Trust – AI can reduce user anxiety, especially among low‑literacy populations, by providing reliable assistance.
  3. Financial Literacy – AI can help users understand complex financial products, fostering informed decision‑making.

4.2 Safia Hussain – Bridging Builders & Beneficiaries (Karya)

• Framed low‑income, rural people as both AI builders and beneficiaries.
• Warned against the “builder vs. beneficiary” dichotomy, which hinders trust and power sharing.
• Stressed the need for language‑ and culture‑sensitive AI; models must be useful in local contexts, not just technically correct.

4.3 Surya Ganguly – Public‑Sector AI Investment & “CERN for AI”

• Highlighted the concentration of AI power in a handful of corporations.
• Argued that open scientific publishing (e.g., the “Attention is All You Need” paper) catalysed breakthroughs like AlphaFold.
• Proposed a large‑scale public‑funded AI research hub (analogy: CERN for AI) to democratise AI knowledge and mitigate inequality.

4.4 Robert Opp – UNDP Perspective on AI & SDGs

• Identified inequality as the defining character of global development.
• Stressed that AI’s potential for accelerated economic and social progress must be inclusive.
• Emphasised investment in ecosystem building: data governance, institutional capacity, and trust‑building.
• Asserted that inclusion builds trust, which is essential to close the equity gap.

4.5 Claire Melamed – Policy Incentives, Investments, and Interoperability

• Warned that AI adoption will inevitably create inequality, but the goal is to prevent systemic exclusion.

• Outlined three levers:

  1. Incentives – Align policy, regulation, and global goals to encourage equitable AI development.
  2. Investment – Public‑ and private‑sector funding must target structural disparities.
  3. Interoperability – Ensure pilots and small‑scale solutions interconnect to scale impact.

4.6 Osama Manzar – Civil‑Society Concerns & Safeguards

• Described the historical pattern: new tech (internet, social media) initially promised empowerment but later exacerbated disintegration.
• Listed AI’s “virtues” (i.e., risks) such as discrimination, hallucination, environmental degradation, bias, algorithmic manipulation, etc.
• Called for robust safeguards and a social‑system‑centred approach to AI deployment.

4.7 Paula Bogantes Zamora – Government‑Level Challenges (Costa Rica)

• Highlighted infrastructure gaps: limited 5G rollout, English‑centric software, and poor data digitisation.
• Discussed AI‑driven job displacement as a combined economic, social, and fiscal challenge.
• Cited Costa Rica’s demographic trends (life‑expectancy up, fertility down) that stress pension systems.
• Proposed shorter certification cycles and rapid reskilling to keep displaced workers employable.

4.8 Closing Reflections (moderator & Amandeep)

• Re‑emphasised that AI diffusion cannot be a “trickle‑down” phenomenon; it must transform systems at scale.
• Summarised the need for political will, infrastructure, local ecosystems, linguistic/cultural relevance, financial literacy, institutional capacity, and human‑centred design.
• Noted upcoming global dialogues (ITU AI for Good, World Summit on the Information Society) where these themes will continue.

---

5. Key Announcements & Calls to Action

Announcement	Speaker	Summary
100 diffusion pathways by 2030	Shankar Maruwada	Collaborative target across sectors & continents to accelerate AI‑driven development.
Public‑funded “CERN for AI”	Surya Ganguly	Proposal for a globally coordinated, publicly financed AI research hub.
Subsidised compute & open‑model support	Kamal Kishore	Policy levers for governments to democratise AI access.
Financial‑inclusion AI roadmap	Bipin Prit Singh	Three‑pillar strategy (native language UI, trust, literacy).
Rights‑based AI framework	Vilas Dhar	Call for norms, principles, values to guide AI development.
Interoperability requirement	Claire Melamed	Need to ensure pilot projects connect into a cohesive ecosystem.

Key Takeaways

• AI must reinforce, not replace, existing disaster‑risk governance; misinformation is a major hazard.
• Open‑source, lightweight models are essential for affordable, inclusive AI deployment, especially in emerging economies.
• “Small AI”—practical, locally‑relevant tools—can deliver immediate benefits while broader infrastructure catches up.
• Diffusion pathways (rapid roll‑outs of AI advisory services) can accelerate sectoral transformation; a target of 100 pathways by 2030 was announced.
• Public‑sector investment (e.g., a “CERN for AI”) is crucial to counterbalance the concentration of AI power in a few corporations.
• Financial inclusion hinges on AI‑enabled native‑language interfaces, trust‑building, and enhanced financial literacy.
• Bridging builders and beneficiaries eliminates a key trust gap; AI must be culturally and linguistically contextualised.
• Interoperability and ecosystem building (data governance, institutional capacity) are prerequisites for scaling AI impact.
• Policy incentives, public‑private funding, and democratic oversight are needed to prevent systemic exclusion and ensure equitable AI benefits.
• Government challenges—infrastructure, multilingual software, data digitisation, and workforce reskilling—must be addressed to harness AI for sustainable development.

---

Prepared from the verbatim transcript of the AI for Development panel at the Delhi AI Conference (2026).

See Also:

• thriving-with-ai-human-potential-skills-and-opportunity
• building-resilient-sustainable-ai-infrastructure-for-people-planet-and-progress
• responsible-ai-at-scale-governance-integrity-and-cyber-readiness-for-a-changing-world
• democratizing-ai-resources-and-building-inclusive-ai-solutions-for-india
• ai-for-democracy-reimagining-governance-in-the-age-of-intelligence