India showed how AI can work at scale, but inclusion remains a challenge: World Bank | Special india news

Last updated:February 22, 2026, 20:51 IST

The World Bank said it is prioritizing “small AI” solutions that are affordable, practical and effective, even where connectivity and infrastructure is limited.

The ‘India AI Impact Summit 2026’ brought together policymakers, industry leaders, multilateral institutions and technologists on an unprecedented scale, signaling a shift in the global conversation around artificial intelligence.

Moving beyond hype and model size, the summit focused on how artificial intelligence (AI) can deliver real-world development outcomes from jobs and productivity to public service delivery, while confronting risks around inequality, exclusion and trust.

For the World Bank Group, the summit was an important moment to advance its vision of AI as a tool for inclusive development. As governments in the global South race to incorporate AI into welfare systems, education, healthcare and governance, the World Bank has positioned itself at the center of debates on responsible adoption, digital public infrastructure, cybersecurity and global safeguards. Its emphasis on “small AI”, practical, affordable systems that work in low-resource settings, reflects a broader effort to ensure that AI development reduces rather than widens the gap.

cnn-news18 spoke to Paul Prossey, Acting Country Director for India at the World Bank, and Mahesh Uttamchandani, Regional Practice Director for Digital and AI in East Asia and the Pacific and South Asia. The conversation ranged from the risks of AI-led exclusion and algorithmic bias to India’s role in shaping global AI norms, the governance challenges of deploying AI at the state level, and why policymakers still avoid uncomfortable truths when it comes to AI and inequality.

Excerpts from the interview:

The World Bank is increasingly touting AI as a development tool, but many argue that it risks increasing inequality in states with less capacity. How do you ensure that AI projects supported by the World Bank do not benefit governments and vendors more than vulnerable populations?

Mahesh Uttamchandani: At the World Bank Group, our focus is clear – AI should promote inclusion, not deepen division. This means designing AI that works for marginalized people, not just for governments or tech vendors. We are prioritizing what we call “small AI” solutions that are affordable, practical, and effective, even where connectivity and infrastructure is limited.

In Andhra Pradesh and Telangana, we are working with governments and partners to assess AI-powered learning tools that help students build job-ready skills. In Uttar Pradesh, AI tools are helping farmers reach wider markets, increase income and create new employment opportunities. These initiatives show that when AI is based on local realities, it can deliver immediate benefits in health, education and agriculture and directly strengthen communities rather than sidelining them.

India’s digital public infrastructure is often held up as a global model. As AI becomes embedded in welfare delivery, health and education systems, what specific risks of exclusion or error most concern the World Bank in the Indian context?

Paul Prossey: India has emerged as a global benchmark for digital public infrastructure. Platforms like Aadhaar and Unified Payments Interface (UPI) show how technology can deliver services at scale with speed and transparency. But as AI becomes embedded in welfare delivery, health and education, new risks come into focus.

The biggest concern is exclusion by design. Algorithmic bias, weak local-language data, or systems trained on non-representative datasets may inadvertently lock out certain communities. There are also serious cyber security risks. Attacks on AI-enabled systems could disrupt essential services or expose sensitive personal data, reducing public trust.

For the World Bank Group, the priority is to put responsible AI governance and cybersecurity at the core, not as an afterthought. This means strong data governance, transparency in the way algorithms are deployed, effective grievance redressal mechanisms and clear lines of accountability.

India has already taken important steps in this direction. The Digital Personal Data Protection Act establishes clear rules on consent, data management responsibilities and cross-border data sharing. Furthering this emphasis on trust, Prime Minister Narendra Modi called for a “glass box” approach to AI at the AI ​​Impact Summit. The idea is simple but powerful – AI systems should be open, explainable, and governed by visible and verifiable security rules, not hidden behind opaque black boxes.

Many Indian states are now experimenting with AI in policing, education and social services. Is the World Bank engaging directly with state governments on AI deployment and, if so, how does it ensure continuity with national and global safeguards?

Paul Prossey: AI governance cannot stop at state or national boundaries. Data flows freely across jurisdictions, and risks such as cyber threats or misinformation do not respect borders. That is why regulation must be rooted in local realities, but based on shared global principles.

The World Bank Group takes a layered approach. At the state level, AI deployment must comply with national laws. Also, it should reflect global best practices on fairness, transparency, accountability and data privacy.

States need space to tailor AI tools to local needs, but within a common safeguards framework. The approach we advocate is risk-based, principles-driven and tailored to each country’s institutional capacity and level of digital maturity.

This philosophy is spread globally. For example, we have supported the development of the African Union AI Continental Strategy, which strikes a balance between regional coordination and national flexibility.

AI may be borderless, but governance cannot be one-dimensional. To ensure that innovation moves forward safely, inclusively and with shared standards of trust, it must operate simultaneously at local, national and global levels.

India is increasingly establishing itself as the voice of the Global South on technology governance. Does the World Bank see India as a co-architect of global AI norms, or primarily as a test case whose lessons are later exported elsewhere?

Paul Prossey: India is both a co-architect of global AI norms and a proving ground for inclusive AI at scale. Its strength lies in its ability to drive innovation through regulatory sandboxes and targeted programs and then rapidly scale up what works. The transition from proof of concept to nationwide impact offers powerful lessons for other developing economies.

More importantly, India is helping reshape the global AI conversation. Instead of focusing only on bigger models or more computing power, it is shifting the debate towards development outcomes such as job creation, productivity gains and better public service delivery. The World Bank Group is partnering with India in this transformation by supporting “small AI”: task-specific, multilingual systems that operate on low bandwidth and basic smartphones.

India’s leadership also matters at the regional and global south levels. Not every country can build large-scale computing infrastructure on its own, but shared facilities, common standards, and open-source partnerships can expand collective capacity. With its scale, technological talent and policy ambition, India is shaping how AI governance and digital development evolves in the global South.

Finally, after listening to leaders and industry voices at the AI ​​Impact Summit, what are the most promising signs you see for AI-led growth – and what is the most inconvenient truth about AI and inequality that policymakers still prefer to avoid?

Mahesh Uttamchandani: The most promising sign is the growing recognition that AI can create jobs and expand opportunity when it is designed for inclusion. “Miniature AI”, practical and affordable tools, are already showing results. We see students receiving personalized learning support, farmers accessing better advisory services, small entrepreneurs building digital credit histories, and clinics providing care to underserved communities. These applications increase productivity and open new avenues of employment for those who are often left behind.

To turn this potential into real jobs and opportunities, countries need to learn from each other. That’s why the World Bank Group, together with six other multilateral development banks, has launched the AI ​​Repository. It brings together real-world AI applications in development, allowing governments to adapt, replicate and scale what is proven to work.

The inconvenient truth is that inclusion also increases risk. As the poorest and most vulnerable are brought into digital systems, risks ranging from fraud and misinformation to algorithmic bias are inevitable. Policy makers still consider security measures as secondary. That is a mistake. Responsible regulation, strong consumer protection, transparency, accountability and human oversight must be built in from the beginning. AI can be a powerful force for development, but managing its risks is not optional. This is a shared responsibility.

read more