One Billion Futures: AI and the Future of Education Equity in the Global South

Abstract

The session examined how responsible AI can become a catalyst for narrowing the widening learning and digital divides that affect the Global South, where 85 % of the world’s youth reside. The panel highlighted stark statistics on out‑of‑school children, gender gaps, and chronic under‑investment in education, then explored concrete ways AI‑driven tools can restore agency, improve equity, and generate scalable learning experiences. Speakers shared experiences from open‑source textbook platforms, AI‑enhanced tutoring systems, and community‑led design, while also flagging infrastructural, linguistic, and ethical challenges. The discussion concluded with a short collaborative sprint that invited audience members to align around four stakeholder personas—parents & communities, policy makers, teachers, and youth—to surface actionable ideas for AI‑enabled education equity.

Detailed Summary

• Khushboo Awasthi opened with a vivid metaphor, comparing technology to the oil that makes Indian “bhajaya” possible, underscoring that code and AI are now inseparable from how we learn, live, and work.
• She emphasized that the challenges confronting the world—geopolitical, socioeconomic, humanitarian—must be tackled collectively; otherwise, AI‑driven solutions will fall short.
• The moderator positioned the conversation at the intersection of education, technology and justice, framing the panel’s mission: re‑imagine AI not merely as an innovation engine but as a force for education equity that nurtures “one billion futures” rather than serving only privileged minorities.

2. Data‑Driven Context: The Global South Landscape

Metric	Figure
Global youth residing in the Global South	> 90 %
Children out of school (Global South)	≈ 270 million
Kids < 10 years who cannot read grade‑appropriate text	> 70 %
Youth not in education, employment or training	1 in 5 (higher for girls)
Average per‑child education expenditure (Global South)	< $55 (≈ 5,500 INR)
External education funding trend	Declining

• Khushboo highlighted the demographic dividend at risk because education equity — the very foundation of that dividend — is eroding.
• She also noted the colonial legacy that still shapes curricula, language of instruction, and chronic under‑investment, setting the stage for why AI must be purpose‑built for local contexts.

3. The Promise and Peril of AI

• AI can amplify voices that were previously unheard; Khushboo shared an anecdote of women in remote Indian villages feeling empowered by an AI‑based companion.
• The panel’s central question: How can AI be deliberately designed to restore agency and dignity rather than deepen existing inequities?

4. Introducing the Panel

• Aditya Vishwanath (Makerghat) was invited to introduce the speakers.
• The panel comprised two women leaders (Neeru Khosla, Vandana Sikka) and two technical leads from CK‑12 (Neeru and Miral Shah), along with the UNESCO chief (Dr. Mishra) and the moderator (Khushboo).

5. Theme 1 – AI’s Evolution for Learning Resources (CK‑12 Perspective)

5.1 Historical Snapshot

• Neeru Khosla recounted CK‑12’s two‑decade journey: initially textbook‑centric with limited online presence. By 2020 they had moved beyond static PDFs to multimodal content, simulations, and adaptive practice systems.

5​.2 The AI‑Enabled “Flywheel”

• CK‑12 now integrates a practice system, knowledge‑tracing, and conversational AI that delivers real‑time diagnostics of a learner’s knowledge state.
• This shift from binary right‑or‑wrong grading to deep, moment‑by‑moment insight enables teachers to intervene precisely when a concept (e.g., fractions, combustion vs. nuclear fission) is misunderstood.

5.3 Personalisation at Scale

• The platform’s Flexbooks (flexible textbooks) were introduced 20 years ago as an early form of personalisation—today’s AI amplifies this by offering step‑by‑step hints, contextual prompts, and adaptive pathways without charging fees.
• CK‑12’s human‑in‑the‑loop workflow curtails hallucinations, ensuring content quality while keeping the service free and non‑commercial.

5.4 Key Insight

“AI should not be a consumer‑grade chatbot but an educational partner that works hand‑in‑hand with teachers, delivering granular feedback to each student.”

6. Theme 2 – AI Literacy Across Divergent Contexts

6.1 Ladder Metaphor

• Dr. Shitanshu Mishra framed AI literacy as a ladder, where the starting rung depends on a child’s existing access.
  • Chhattisgarh (low‑connectivity): focus on misinformation detection, safe query formulation, and voice‑based or WhatsApp chat‑bot tutors.
  • Palo Alto (high‑connectivity): deeper probes into model bias, data‑sets, simple model‑building, and career pathways.

6.2 Foundational vs. Advanced Literacy

• Mishra cautioned against leap‑frogging foundational literacy (reading, numeracy). AI literacy must be scaffolded on the same solid base, mirroring strategies used in Code.org’s CS curriculum and Pratham’s grassroots teaching.

6.3 Multi‑Stakeholder Blueprint

Stakeholder	Role in AI Literacy Development
Local Communities	Define contextual learning objectives
Governments	Provide infrastructure, set guardrails
Educators	Design pedagogy, adapt tools for classroom
Tech Companies	Offer transparent tools, free or low‑cost access

• Mishra stressed that AI literacy must be a partnership—the goal is not merely to create consumers of AI, but “AI shapers” who can build and guide future systems.

7. Theme 3 – Infrastructure, Cost, and Energy Constraints

7.1 Dual‑Layer Infrastructure

• Aditya Vishwanath outlined two infrastructure layers:
  1. End‑user layer – broadband, electricity, devices.
  2. Model‑training layer – GPU clusters, data‑centers, energy consumption.

7.2 Offline‑First Design

• Recommendations included offline‑first architectures (e.g., WhatsApp chatbots, voice‑enabled tools) that function in low‑connectivity environments.
• Existing successful models cited: Pratham’s Padhai & Bal Siksha, Khan Academy’s offline resources, and CK‑12’s Flexbooks.

7.3 Cost Disparities

• Reality check: India’s AI ecosystem currently has ≈ 50,000 NVIDIA GPUs, aiming for 200,000 by year‑end, yet OpenAI accessed > 1 million GPUs last year; Africa has < 10,000.
• Energy demand: a 3 GW data centre under construction in India illustrates the massive electricity footprint of AI training.

7.4 Opportunity for Local Talent

• The shortage of high‑end compute resources creates a demand for local expertise in energy management, hardware maintenance, and AI model optimisation—areas where Indian talent could lead.

8. Theme 4 – Contextualising AI for Linguistic & Cultural Diversity

• Mishra highlighted 22 Indian languages spoken by ≥ 100 million speakers each.
• The panel discussed how CK‑12 ensures language‑agnostic, culturally relevant curricula by:
  • Building modular content that can be localized without re‑engineering the entire system.
  • Embedding learning‑science principles (e.g., spaced repetition, mastery learning) that are universal, while allowing local teachers to add contextual examples.

9. Theme 5 – Funding, Sustainability, and the Non‑Profit Model

• Vandana Sikka described Learny’s Vaasi app: a free‑to‑download platform that lets anyone create AI‑enhanced knowledge nuggets, reinforcing the principle that “everyone is great at something.”
• CK‑12’s non‑profit stance—no data sales, free content for all—was positioned as a counter‑weight to commercial AI models that may prioritize profit over equity.

10. Interactive Collaborative Sprint

• Khushboo transitioned the session to a 15‑minute “persona sprint.” Four stakeholder personas were identified:

  1. Parents & Communities – led by Ruchi (CEO, Atma Shakti Trust)
  2. Policy Makers & Administrators – led by Dr. Shitanshu Mishra
  3. Teachers & Educators – led by Neeraj (CEO, Prava)
  4. Youth & Children – led by Neeraj (central position)

• Participants were asked to physically move to the corner representing the persona they resonated with, creating a dynamic “standing” brainstorming space.

• The activity was framed as a “jugaad” (resourceful improvisation) to surface concrete ideas from each stakeholder group on how AI can be deployed today, despite infrastructural limits.

11. Closing Remarks & Acknowledgements

• Khushboo thanked the speakers, the audience (≈ 600 concurrent sessions), and emphasized that education is never a one‑way street; success depends on continuous two‑way engagement between innovators and end‑users.
• A final round of applause was prompted for Neeru, Vandana, and the broader panel.

Key Takeaways

• AI as an Equalizer – When purpose‑built and integrated with learning science, AI can restore agency to learners in the Global South, moving beyond the “consumer‑grade” chatbot model.
• Scale of the Challenge – Over 270 million children are out of school; 70 % of kids under ten cannot read at grade level; 1 in 5 youths lack education, employment, or training.
• Infrastructure Duality – Effective AI adoption requires addressing both end‑user connectivity/device gaps and the high‑end compute/energy demands of large language models.
• Offline‑First Solutions – Leveraging WhatsApp chatbots, voice‑assistants, and low‑bandwidth content can deliver AI‑enhanced learning where internet access is spotty.
• AI Literacy Ladder – Literacy must be scaffolded: low‑connectivity contexts start with safe query practices; high‑connectivity contexts can explore model bias and creation.
• Non‑Profit, Open‑Access Models – Platforms like CK‑12 and Learny’s Vaasi demonstrate that free, open, human‑in‑the‑loop AI can maintain quality while staying accessible.
• Cultural & Linguistic Localization – AI tools need modular, language‑agnostic design to serve India’s 22 major languages and other Global South contexts.
• Stakeholder Partnerships – Sustainable AI‑enabled education demands co‑design among communities, governments, educators, and tech providers.
• Talent & Energy Opportunities – The need for local GPU infrastructure and energy‑efficient AI creates emerging career paths in hardware, renewable energy, and AI engineering.
• Interactive Persona Sprint – Engaging participants through physical movement and persona‑based brainstorming can quickly surface actionable, context‑specific ideas, embodying the “jugaad” spirit of frugal innovation.

---

See Also:

• flipping-the-script-how-the-global-majority-can-recode-the-ai-economy
• empowering-the-human-edge-building-a-future-ready-workforce-in-the-age-of-ai
• ai-for-economic-growth-and-social-good-ai-for-all-driving-economic-advancement-and-societal-well-being
• reducing-ai-driven-digital-harms-for-minors-platform-accountability-age-assurance-and-safety-by-design