Introduction

The advent of generative artificial intelligence (AI) holds the potential to revolutionize the way public digital services are designed, delivered, and consumed. The ultimate goal is to improve the quality of digital services provided by public administrations to citizens, leveraging AI's potential to analyze large volumes of data, generate innovative solutions, and optimize decision-making processes while ensuring respect for the constitutional principle of good governance, data protection, transparency, accountability, inclusiveness, and security.

Understanding Generative AI

Generative AI is a branch of artificial intelligence that focuses on creating algorithms capable of producing original and variable outputs based on existing data and information. The generation can involve text, images, sounds, videos, or other forms of content. Generative AI relies on machine learning techniques such as generative adversarial networks (GANs) and large language models (LLMs) to create new content and solutions.

Generative AI must be distinguished from two complementary concepts:

1. Artificial General Intelligence (AGI): AGI, or strong AI, refers to a type of artificial intelligence that possesses the ability to understand, learn, and apply knowledge across a wide range of tasks and disciplines, as a human would. AGI remains a theoretical objective, partly because there is no universally accepted definition of human intelligence. Generative AI is a more specific and limited application of artificial intelligence and does not reach the level of competence and adaptability of AGI.
2. General Purpose Artificial Intelligence (GPAI): General-purpose AI refers to AI systems and algorithms that can be applied to a wide range of tasks and problems without being specifically designed for a particular application. An example of general-purpose AI is transformation language models like GPT-4, which can be used in various contexts such as text generation, automatic translation, text summarization, and question answering. Generative AI can be considered a form of general-purpose AI since it can generate content in various formats and contexts.

Generative AI and Public Administration: A Design Challenge

The integration of generative AI into public digital services is primarily a design challenge. Recent years have seen an increased focus on design within public digital services in Italy, driven by both regulatory and practical commitments to guide digital transformation in public administration.

The Agency for Digital Italy, in collaboration with the Department for Digital Transformation of the Presidency of the Council, has adopted guidelines for the design of websites and digital services for public administrations. These guidelines, enriched by public consultation and dialogue with relevant organizations, aim to provide operational guidance in designing and implementing digital touchpoints for citizens.

A conscious, open, and citizen-centered design culture is essential for tackling the challenge of digitalization in the public sector. Design can ensure that public administration innovation occurs not only from a strictly technological standpoint but also as an opportunity to re-examine and improve underlying processes, adopting a tailored approach for the public context.

Key Principles for Integrating Generative AI in Public Digital Services Design

While explicit reference to generative AI may be absent in the mentioned guidelines, some key principles remain valid for integrating this technology into the design of public digital services:

1. "Once-only" principle: The once-only principle, a key concept in the design of public digital services, states that information provided by citizens or businesses to public administrations should be collected only once and then shared and reused among different government entities, reducing the need for users to repeatedly provide the same information.
2. User-centered design: Digital public services must be designed considering the needs, expectations, and preferences of end-users. Involving users in all stages of the design and development process is essential for creating solutions that meet their needs and are easy to use.
3. Accessibility and inclusiveness: The design of digital public services with generative AI must ensure that they are accessible and inclusive for all citizens, regardless of their abilities, technological knowledge (digital literacy), or socio-economic conditions. This implies adopting accessibility standards and designing interfaces and functionalities that enable all users to access and use the services effectively.
4. Transparency and communication: Design should facilitate users' understanding of how generative AI functions and the underlying decision-making process. This involves creating interfaces that clearly and simply explain AI decisions, the criteria used, and the source of information (although the explainability of AI, in general, is a complex issue).
5. Security and data protection: Adequate design must ensure that digital public services with generative AI protect users' personal data and comply with privacy regulations, through the adoption of appropriate security measures and clear communication of policies on data collection, use, and storage.
6. Flexibility and scalability: The design of digital public services with generative AI must foresee the possibility of adapting to new needs, technologies, and contexts. This implies creating modular and scalable solutions that allow for changes and improvements over time without compromising the user experience.

Conclusion

Integrating generative AI into the design of public digital services offers immense potential to enhance the quality and efficiency of these services. By adhering to key principles such as the once-only principle, user-centered design, accessibility, transparency, security, and flexibility, public administrations can leverage generative AI to create innovative and user-friendly solutions that benefit citizens and businesses alike.