The State of Gov.ai: AI for Public Good
Artificial intelligence (AI) is evolving rapidly, with core metrics such as model parameters, compute capacity, and investment flows expanding significantly. Yet, the development of AI applications aimed at public benefit is not receiving adequate attention. While AI is often perceived as a tool for cost reduction and enhanced service delivery, only approximately 25% of government agencies have adopted AI in any substantial form. As AI continues to shape global economies and industries, its most transformative potential may reside in its use for public purposes rather than private profits.
By 2025, over 70 nations are expected to have established national AI strategies, many of which will incorporate AI into the public sector. However, actual implementation is trailing behind these intentions. Common barriers include poor data quality, inadequate digital infrastructure, privacy and security concerns, high investment requirements, and the lack of AI frameworks that align with public sector needs.
This five-part series advocates for a novel framework to develop AI for public good, focusing particularly on enhancing service delivery. The objective is not just to keep pace with AI innovation but also to ensure that its advantages are equitably and effectively shared.
The Opportunity: Combining DPI with AI
We find ourselves in a pivotal moment where Digital Public Infrastructure (DPI) is being harnessed to revolutionize public service delivery on a large scale. DPI is characterized by its core principles—public good, openness, transparency, interoperability, user control, privacy-by-design, and a comprehensive societal approach.
The fusion of AI with the public-oriented architecture of DPI can enhance the provision of public services, making them more effective, just, and transparent. An AI framework that incorporates DPI’s fundamental attributes can facilitate a significant shift in how public services are delivered.
Digital Public Agents: Governance in the Intelligent Age
The future of AI is anticipated to be dominated by Agentic AI. This new paradigm, termed ReAct, signifies a transition from simple reasoning to actionable outcomes. AI Agents will play essential roles across both consumer and enterprise applications.
Given the adaptability and capabilities of AI Agents, they serve as logical interfaces for integrating DPI into AI systems. This leads to the conception of a new type of AI Agent—the Digital Public Agent (DPA).
A Digital Public Agent is defined as:
“a new building block in digital transformation that combines the core attributes of DPI—public good, openness, transparency, interoperability, user control, and privacy-by-design—with the cognitive, analytical, predictive, reasoning, and executive powers of AI, to benefit society.”
In essence, DPAs are typical AI Agents augmented with features necessary for carrying out public functions and interacting with governmental entities.
DPAs can perform foundational tasks such as establishing individuals’ or businesses’ identities, facilitating payments, and gathering information or feedback from government agencies. They can also evolve into composite AI agents capable of registering individuals or businesses with statutory agencies, filing returns, and delivering public services while monitoring public programs.
DPAs have the potential to transform service delivery by providing context-sensitive, hyper-personalized services and breaking down silos within government agencies. Event-driven DPAs can react to significant life or business events—such as births, school admissions, marriages, home constructions, job changes, incorporations, audits, or leadership transitions—thus adding timely and relevant value.
DPAs can be designed with the following characteristics:
- DPAs can target either the consumer (citizens or businesses) or the service provider (predominantly government).
- They can function as foundational or sector-agnostic entities, such as proving one’s identity and processing payments, or they can be sector-specific in areas like tax returns and granting or applying for permits or licenses.
- DPAs can be engineered to enhance productivity, ensure compliance, and monitor performance and operations.
In summary, DPAs have the potential to serve as skilled assistants to both citizens and businesses, as well as to public sector employees.
Open Issues
While significant claims have been made regarding the capabilities of DPAs, several key questions remain, including:
- How can government processes be restructured to be ‘fit-for-DPA’?
- In what ways can DPAs be designed to promote public good while leveraging AI effectively?
- What are typical use cases for DPAs?
- What are the challenges and opportunities in realizing the DPA concept?
These questions will be addressed in the forthcoming articles in this series.
The author of this article is Mr. J Satyanarayana, a former member of the Indian Administrative Service and the former IT Secretary for the Government of India and Chairman of UIDAI (Aadhaar).
Disclaimer: The views expressed are solely those of the author, and ETCIO does not necessarily endorse them, nor is it responsible for any potential consequences arising from them.
Note: This article is the first installment of a five-part series on Digital Public Agents.
