Inclusive AI in India: Making Technology Accessible to Everyone

Contents

Executive Summary

This session at a global AI summit explored how India can democratize artificial intelligence by making it accessible, affordable, and inclusive across all socioeconomic strata and linguistic communities. Speakers emphasized that democratizing AI requires far more than technological innovation—it demands coordinated action across governance, infrastructure, skill development, and responsible deployment. India is uniquely positioned to lead this effort by leveraging its experience with digital public goods (Aadhaar, UPI, CoWin) and extending inclusive digital infrastructure into the AI era.

Key Takeaways

AI democratization requires holistic approach: Technology alone is insufficient. Success demands coordinated advancement in governance frameworks, infrastructure investment, skill development, trustworthy data, and responsible deployment—not technology first.
India can model inclusive AI for the world: India's experience building scalable digital public goods (Aadhaar, UPI, CoWin) offers replicable pathways for other developing nations facing similar challenges. India can demonstrate that AI need not concentrate power; it can distribute opportunity.
Language and data are non-negotiable: For AI to truly serve Indians, it must operate in Indian languages and be trained on India-specific contexts. Relying solely on English-language, Western-trained models will perpetuate digital colonialism and undermine inclusivity.
Edge computing and privacy-by-design preserve sovereignty: Bringing computation closer to users—rather than centralizing all processing in distant data centers—protects privacy, enables offline functionality, and preserves data sovereignty while delivering AI benefits.
The next phase is enabling contributors, not just users: True democratization means not only giving people access to AI but enabling them to contribute data, verify information, and shape AI systems. Community-driven models (Wikipedia, open knowledge platforms) are essential mechanisms for this.

Key Topics Covered

Democratization of AI as a developmental imperative for the Global South and India's inclusive growth model
Digital public infrastructure as a model for AI accessibility (Aadhaar, UPI, CoWin examples)
Multilingual and language inclusion in AI systems across India's linguistic diversity
Skill gap and workforce reskilling initiatives to prepare India's population for AI-driven economy
Data sovereignty and high-quality datasets — training models on India-specific contexts rather than Western-centric data
Computing accessibility across tier 1, 2, 3 cities and rural areas through hybrid/edge computing
Financial inclusion and fraud prevention using AI-powered solutions (e.g., UPI security)
Open knowledge ecosystems and community-driven platforms (Wikipedia) for verified, trustworthy datasets
Responsible and ethical AI governance frameworks balancing innovation with safeguards
Startup and SME enablement through cost-effective AI solutions and access to capital

Key Points & Insights

AI is infrastructure, not luxury: AI is fast becoming foundational infrastructure like electricity and the internet. If concentrated in few hands, it will widen inequalities between nations, regions, and citizens. India must ensure AI becomes a public resource empowering farmers, teachers, entrepreneurs, and health workers in underserved areas.
Data quality over quantity: The biggest currency in AI is high-quality, contextually relevant data. Most large language models are trained on Western datasets. India must build or train models on India-specific data (language, culture, regional context) to ensure models serve Indian populations effectively. Low-quality training data produces deceptive, realistic but fundamentally flawed outputs.
Government's AI infrastructure role: India's AI Mission is creating shared compute infrastructure to remove cost barriers for startups, academics, and innovators. High-quality anonymized, consent-based public datasets are being prioritized. Talent development is being expanded beyond elite institutions to tier 2 and tier 3 cities.
Language inclusion as dignity and empowerment: India has hundreds of languages and dialects. AI systems must understand this linguistic diversity to serve citizens with dignity. Initiatives like Bhashini are enabling speech and language technologies across Indian languages, allowing citizens to interact with digital services in their mother tongue—not merely convenience, but a human rights issue.
AI solves problems it creates: AI-powered fraud detection in UPI prevents ~₹13,000 crores (≈$1.5B) in fraud annually. AI credit models are enabling financial inclusion by detecting creditworthiness signals from digital payment trails rather than traditional banking data, giving entrepreneurs access to capital previously unavailable.
Hybrid and edge computing for sovereignty: Not all computing needs to happen in centralized data centers. Hybrid LLM models, edge computing, and local inferencing keep user data private and preserve data sovereignty while delivering AI benefits. Intel and others are working on technologies to bring AI compute closer to the user.
Community-driven knowledge ecosystems matter: Wikipedia and open-knowledge platforms provide verified, source-backed content for training datasets—crucial where Indic language datasets represent <1% of available training data. Community verification ensures trustworthiness; expanding contributor communities across India is critical to filling data gaps.
Inclusive design requires multistakeholder participation: Democratizing AI cannot be achieved by government alone. Industry must invest in inclusive design and responsible deployment; academia must expand interdisciplinary research rooted in societal needs; startups must innovate for local markets; civil society must ensure accountability; international partnerships must enable equitable access.
Skills development as a prerequisite: Despite opportunities, the skills gap remains critical. Universities are integrating AI into all disciplines (agriculture, health, arts, management), not just engineering. Intel pledged to train 100,000+ users in AI during the summit. Skilling at scale is essential—people need to be trained to use and contribute to AI systems.
Optimism over fear in AI adoption: While legitimate concerns exist (bias, job displacement, monopolistic concentration), the speaker from Google Pay argued for an optimistic approach: "If we approach the problem from optimism, there are problems we can solve in interesting ways." The choice between building AI as a weapon or tool of mass transit mirrors historical technology choices.

Notable Quotes or Statements

"Who is AI really for? Is it for those with access to vast computing power and proprietary datasets... or can AI become a public resource something that empowers a farmer in Bundelkhand or a school teacher in Northeast or a small entrepreneur in Coimbatore?" — Mahavir Singh (Additional Secretary, Ministry of External Affairs)

"If AI remains concentrated in few hands, it will deepen divides. If AI becomes accessible, inclusive and trusted, it can become the greatest force multiplier for human progress." — Mahavir Singh

"Imagine you train a child with foul language. It will be the most drastic thing that can happen. If AI is trained with wrong data, the outcome will be so realistic you won't be able to say this is wrong. Data is the biggest currency in the AI world." — Sepenna (Chief Growth Officer, ACLT Tech)

"It's the best of times, it's the worst of times... a time of fear and a time of hope. If we approach the problem from optimism, there are problems we can solve in very interesting ways." — Sharat Bulusu (Senior Director, Google Pay India)

"Language is one of the areas where I'm very proud of the work... there are languages that would otherwise go extinct that we are able to protect through translation and creating corpuses." — Sharat Bulusu

"Democratizing AI is not merely a technological choice. It is also a developmental imperative, particularly for a Global Southern country like India." — Mahavir Singh

Speakers & Organizations Mentioned

Speaker	Role/Title	Organization
Mahavir Singh (referred to as "Mahi G")	Additional Secretary, New Emerging and Strategic Technologies	Ministry of External Affairs, Government of India
Brad Staples	CEO	APCO (Global advisory group)
Sepenna	Chief Growth Officer, India; MEA France; Italy	ACLT Tech (London-based, moving to Mumbai)
Sharat Bulusu	Senior Director	Google Pay India
Samir Fukan	Director, Head of Customer and Partner Engineering	Intel India
Pravin Das	Lead Partnerships Manager, South Asia	Wikimedia Foundation
Rajita Kulkarni (referred to as "Rajita" / "Rajitha")	President	Shishu University
Session Moderator		(Appears to be opening remarks speaker, not fully identified)

Other organizations/initiatives referenced:

Art of Living Foundation (parent organization of Shishu University; present in 184 countries)
IBM, Verics (AI partnerships with Shishu University)
Institute of Absolute Intelligence / Foundation for Artificial Intelligence
Department of Science and Technology, Government of India
Wikipedia, Wikimedia Foundation

Technical Concepts & Resources

Concept/Initiative	Description
Bhashini	Government initiative enabling speech and language technologies across Indian languages
India AI Mission	Government program creating shared AI compute infrastructure, building public datasets, and funding AI skilling initiatives
UPI (Unified Payments Interface)	Digital payment platform; mentioned as democratizing financial access; 12 billion transactions/month; serves as a model for digital public infrastructure
Aadhaar	Digital identity system; provided identity to 1.4 billion people; cited as example of inclusive digital public infrastructure
CoWin	Vaccination management system; enabled world's largest vaccination drive; example of inclusive digital platform
Hybrid LLMs	Language model architecture where inference and partial training occur locally/at the edge, with references back to centralized systems only when necessary
Edge Computing	Bringing computation closer to users rather than centralizing in data centers; preserves privacy and data sovereignty
Large Language Models (LLMs)	Currently trained predominantly on Western/English data; must be adapted/retrained on India-specific datasets
Photonix	Intel technology for improving connectivity (mentioned in context of system-level solutions)
Computer Express	Intel technology for improving connectivity
AI-powered fraud detection in UPI	System preventing ~~₹13,000 crores (~~$1.5 billion USD) in fraud annually
Data Commons for Low-Resource Languages	Initiative to build trustworthy, verified datasets for languages with limited existing data
ChatGPT Plugin (Wikipedia)	Integration to ensure AI model users access trustworthy sources when querying
NDP (National Data Portal or similar)	Framework enabling industry-integrated programs in higher education
AI for All program	Foundational AI literacy program for all academic disciplines (partnership between Intel, Shishu University, and others)

Data and metrics:

Indic language datasets represent <1% of total LLM training data
UPI moves >₹4 trillion annually
India has 900 million internet subscribers (with stated need for further accessibility expansion)
Shishu University incubation: 180+ AI-based startups (in partnership with Department of Science and Technology)
Intel AI pledge during summit: 100,000+ users trained in AI (with indication this target was exceeded)

Additional Context

The session occurred at a global AI summit and was broadcast live to online audiences. The discussion emphasized that India's approach combines open innovation with digital sovereignty—promoting open ecosystems and global collaboration while protecting national data security, privacy, and strategic autonomy. The panelists consistently framed democratizing AI not as a burden but as an opportunity for India to lead a new global paradigm where AI is "an instrument of inclusion, a platform for empowerment, not a luxury for the few but a utility for the many, and not merely intelligent but humane."