Naveen Varshan Ilavarasan, DPI Specialist, UNDP

As countries advance their efforts towards designing and implementing digital systems that can deliver timely, large-scale public and private benefits, artificial intelligence (AI) and digital public infrastructure (DPI) have emerged as two critical enablers. Recognizing this potential, the G20 Troika of Brazil, India and South Africa — three major developing economies — emphasized that well-designed DPI, when augmented by AI, can transform lives and accelerate public outcomes. Though conceptually different, the intersections between AI and DPI are particularly evident through three key dimensions. 

1. The foundational linkage between DPI and AI, where DPI provides the backbone that makes AI effective, scalable and relevant for people.
2. The real-world applications that demonstrate how AI and DPI together serve public needs across sectors.
3. The principles that help guide AI integration across DPI, underscoring the need for safeguards to ensure value for all.

In recent months, this evolving intersection has gained momentum, not just as a technological opportunity, but as a strategic importance. Nandan Nilekani, former Chairman of India’s Unique Identification Authority, has emphasized the potential of combining foundational digital public infrastructure with AI to drive scale, inclusion and public value. Others, including UCL’s Professor David Eaves and Sarosh Nagar, have highlighted the distinct trajectories of AI and DPI, as well as the possibilities for their convergence to strengthen public outcomes.

This leadership—alongside the focus of the G20 Troika countries highlighted earlier—has contributed to a growing recognition that aligning AI with DPI offers a powerful pathway to deliver meaningful public value, and that this alignment cannot be left to chance. Across countries where UNDP works, this is not merely a trend, but a deeper transformation that demands country leadership, homegrown innovation and a whole-of-society approach—including government, civil society, the private sector and the public themselves.

At the foundational level, DPI can provide the underlying data infrastructure—such as civil,functional or entity-based registries, non-personal data and open datasets—that enhance the application of AI in real-world contexts. When such high-quality, country-specific and consent-based data related to public services is available, AI models and applications can be trained in ways that reflect local needs and contexts.

Central to DPI is a focus on interoperability, allowing different systems—such as identity and healthcare, or payments and social protection—to connect and work together. This, in turn, unlocks the integration of diverse datasets, enriches the context for AI use, and enables the development of AI-powered services that draw on multiple systems.

Moreover, DPI enables scalable deployment by offering a shared digital foundation—such as authentication capabilities and service delivery networks—that existing public services already rely on, making it a cost-effective and sustainable backbone to deliver AI-powered services to large populations.

Real-world applications across sectors

The availability of open and high-quality datasets enabled by a DPI approach—such as databases of multiple languages (specific to a country) and sector-specific data, such as local crop and weather data—enables AI models to be trained on locally relevant data and support multilingual communication.

These models then power public-facing services: in India, farmers receive AI-powered advisories to support decisions on crop insurance, government schemes and financial assistance, delivered in local languages and accessible through voice queries as well. In Bangladesh and India, AI supports the translation of court judgments, improving access to legal information for the public.

Behind the scenes, AI also enhances how systems operate (currently the common use of AI)—strengthening fraud detection in digital finance platforms or enabling biometric verification in national ID systems—building trust and efficiency within digital infrastructure.

Mainstreaming AI and DPI in national strategies

The broader application of AI across public infrastructure such as healthcare, education and agriculture, however, remains limited, particularly in countries early in their DPI journey. This presents a unique opportunity for countries to design from the outset for more expansive use, grounded in sound governance and safeguards and focused on practical use cases that deliver value to people.

As global momentum builds, institutions are beginning to support both DPI development and national AI approaches—often as parallel efforts. Making these efforts effective, however, requires broad multi-stakeholder participation—for instance, academic institutions contributing critical datasets, private sector actors proactively developing solutions, and government agencies delivering services to the public.

UNDP, for example, is working with numerous countries on DPI initiatives, while also supporting countries such as Cameroon, Serbia and Ecuador in shaping their AI approaches alongside ecosystem partners. These efforts offer a valuable foundation to drive AI and DPI integration to enhance public outcomes. In Sri Lanka, this conversation has started, with UNDP helping to catalyze early dialogue on the intersection of AI and DPI, and looking to support more countries in this journey.

Designing future-ready systems with safeguards

As countries begin to explore AI within their digital infrastructure, it becomes essential to ensure these systems are designed with responsibility and the public interest at their core. The Universal DPI Safeguards Framework provides a strong foundation for this—guiding the design and implementation of DPI, especially as AI is layered onto it.

These principles are also informed by implementation insights from countries around the world—whether it’s Brazil advancing safeguards through multi-stakeholder collaboration, or Finland ensuring secure data exchange across organizations—each helping shape how safeguards are applied in practice. Ensuring appropriate regulations, data protection policies and privacy safeguards will be key to making the DPI and AI intersection work for everyone.

As large language models become increasingly commoditized and DPI continues to take root in more countries, this combination holds the potential to unlock cost-effective, diverse and human-centric solutions. When open, consent-based local data is treated as a foundational layer of infrastructure, it enables the development of AI solutions that are not only technically robust, but also locally meaningful and grounded in public interest—across both public and private sector applications. Designing for this future—deliberately with a whole-of-society approach—will be key to realizing its full potential.