The humanitarian sector stands at the cusp of a technological revolution that could fundamentally transform how we deliver services to the world’s most vulnerable populations. A new research paper from University College London – Interactions Between Artificial Intelligence and Digital Public Infrastructure: Concepts, Benefits, and Challenges -reveals how artificial intelligence and digital public infrastructure (DPI) can work together to create more inclusive, efficient, and equitable public services.
Defining the Game-Changers
- AI is a general-purpose technology (GPT), like electricity before it with broad applications across society.
- DPI refers to the foundational digital systems that enable everyday functions like digital payment, identification, and data exchange like India’s Aadhaar digital ID system.
The magic happens when these technologies intersect. AI’s general-purpose nature means it can enhance virtually any DPI system, while DPI provides the structured, consent-based data that can make AI systems more accurate and inclusive.
Real-World Impact
The most compelling evidence comes from countries already implementing these combinations. India’s Bhashini system exemplifies the transformative potential. This AI-powered translation platform uses natural language processing, a machine learning technique most commonly used with text, to perform rapid translations between Indic languages. The system leverages crowdsourced collaborators nationwide to improve translations, particularly for languages with limited data.
For humanitarian organizations working in multilingual contexts, this represents a breakthrough. Imagine being able to provide real-time, high-quality translation services across dozens of local languages, ensuring that displaced populations or rural communities can access critical services regardless of linguistic barriers.
Singapore offers another powerful example with Singpass, which recently integrated machine learning for user authentication to combat fraud. As the research notes, this helps improve citizens’ trust in the system, which can facilitate broader uptake.
For humanitarian programs dealing with identity verification challenges, such AI-enhanced authentication could dramatically improve both security and user experience.
Denmark’s Muni chatbot, serving 37 municipalities, demonstrates how AI can streamline service delivery. This points toward a future where displaced populations could interact with AI assistants in their own languages to navigate complex bureaucratic processes for everything from asylum applications to healthcare access.
Data Foundation Revolution
Perhaps even more significant is how DPI can serve as the foundation for more inclusive AI systems. The researchers highlight a critical problem: “large language models will consume all human-generated data by 2028,” with synthetic alternatives proving inadequate.
DPI offers a solution through consent-based data collection at massive scale. India’s Aadhaar system, with nearly 1.38 billion users, demonstrates this potential. The Indian government has created public datasets from this data that AI startups can use to develop better systems, essentially an informal version of ‘industrial policy’ for AI.
DPI can help address algorithmic bias. The research explains that certain historically marginalized populations are often underrepresented in AI datasets, resulting in algorithmic bias. However, if DPI systems achieve universal adoption, they could capture data from marginalized communities, including traditional knowledge of Indigenous populations, making AI systems more representative and accurate for these populations.
Confronting the Challenges
The researchers don’t sugarcoat the obstacles. High inference costs pose immediate concerns—when DPI serves hundreds of millions or even billions of citizens, AI integration could strain government budgets. Interoperability challenges with legacy systems create additional barriers, particularly relevant for development contexts where outdated technology is common.
More concerning are the ethical considerations.
The benefits of using DPI data for AI development require inclusive and effective implementation of DPI, or else risk becoming a double-edged sword. If marginalized communities don’t adopt DPI systems due to systemic disparities or distrust, the resulting datasets could actually worsen algorithmic bias.
Privacy represents another critical challenge. As the researchers emphasize, it is vital that all data is collected with individuals’ informed consent before its collection—otherwise, DPI risks infringing on important individual rights for privacy.
The Development Imperative
For humanitarian organizations, these developments demand immediate attention. Countries implementing AI-enhanced DPI will leapfrog others in service delivery capabilities. Organizations that understand and advocate for inclusive AI-DPI integration will be better positioned to serve beneficiaries effectively.
We must push for DPI implementations that prioritize universal access, ensuring marginalized populations aren’t left behind. We need to advocate for transparent, ethical data governance that respects privacy while enabling innovation. Most importantly, we must begin experimenting with AI-enhanced service delivery in our own programs.
The convergence of AI and DPI isn’t just a technological trend—it’s a humanitarian imperative. Done right, it could democratize access to high-quality public services globally. Done wrong, it could deepen existing inequalities. The choice is ours to make, but the window for influence is narrowing rapidly.
