And now? The first steps for an AI-ready PA

This section represents a list of suggestions that is definitely not exhaustive on what could be the first steps that the Public Administrations, also through the support of AgID, can follow to start implementing AI or to fully exploit its potential, without incurring the risks raised in the challenges of this White Paper. Some tools are already used by the administrations that feed them in daily life to increase the efficiency and effectiveness of services; in other cases, it is necessary to think of new ones, precisely to cope with the lack of adequate solutions of Artificial Intelligence that support the administration in the administrative activity and in the management of services. In this paragraph, we aim to indicate a small “toolbox”, a starting point pending more operational steps and paths that can give a more solid and clear context to new technological solutions, sustainable implementation of Artificial Intelligence really at the service of citizens.

Area of application of AI solution

Public administrations and dedicated project teams faced with the issue of the role that AI can play in the management of their activities must move from a starting point: AI is not always a panacea and is not suitable for any type of challenge. Therefore, it is necessary to analyse in advance which internal processes or services to citizens can be facilitated by the use of technologies based on AI and, at that point, develop the expectations and impact in relation to the support that can derive in terms of efficiency and/or fairness.

Well-planned initial investiments

Once the Public Administrations have identified the areas in which AI can really help, it is necessary to reflect on the resources they need, starting with the budget necessary to implement the new technologies, followed by the involvement of experts (external or even internal if appropriately formed) that know the most suitable AI technology for the need expressed. Furthermore, the Administrations cannot take into account the time necessary for the internal structure to implement the AI solutions identified, especially when large quantities of data and information must be uploaded to develop them. The Task Force of the Italian Digital Agency remains available to explore forms of co-financing and identification of the needs of the Administration through the email intelligenza-artificiale@agid.gov.it.

Let’s not forget the three-year plan

AI requires accelerating the path of digital transformation. The Three-Year Plan provides all the indications to build the foundations on which it will be possible to implement AI solutions. There are already many tools useful for the implementation of the Three-Year Plan, such as the minimum measures for ICT security to public administrations, which must be adopted to counter the most common and frequent threats to which their information systems are subject. Or the many procurement channels that can be activated, depending on the financial perimeter and the diversified technical needs: from the more classic appeal to the Electronic Market for Public Administration (MEPA) and to the Public Connection System (SPC), in particular, lot 3 on Interoperability for data and application cooperation and lot 4 on the creation and management of portals, APP, websites and web applications), up to the procurement tools for innovation. In particular, on this last front, the references are those of pre-commercial procurement (PCP, both on a national scale also in the context of regional planning through European funds) and finally to the public procurement of services and innovative products (PPI).

Start small

As with many other technologies, it is advisable to test the AI on a small scale before applying it at full capacity in its activities. Developing a pilot program allows those who decide to implement AI solutions to become familiar with the technology and to correct any errors during development thus allowing the service itself to improve. There are several models that can be used to set up projects in an agile manner: the site designers.italia.it provides a tailor-made kit for the PA for the development and design of projects. In addition, AI involves a series of risks, well highlighted in the various challenges of the White Paper, related to data protection and privacy, ethical dilemmas, the risk of bias deviations. In order to limit current and future problems as much as possible, it is a good practice to involve the entities involved in AI-based services from the pilot project, exposing data and algorithms in a transparent and replicable manner.

Work on data

It is useful to start from the need of AI to feed on data, at the base of every application of these technologies. For this reason, a typical challenge faced by those who use AI is the access, availability and quality of data. The higher the quality of the data, the better the accuracy and performance of the AI system will be. However, public data is often collected by different administrations and, in many cases, it is fragmented, limited and not easily accessible. Thus, all administrations must ensure the quality and usability of the data they provide, so as to be easily employed to test, use and refine AI systems. Useful tools, modelled on the needs of the PA for the use of data, their interpretation and procedures for release, are the site dati.gov.it and the National Guidelines for the Valorisation of Public Information Assets.

Development of professionalism and skills

Universities have great expertise in the AI sector, with dedicated R&D projects and programs, in addition to the new training and teaching offer of courses and the establishment of ad hoc departments. At the same time, administrations need substantial investment in human capital to manage the growth and potential of new technological systems. Universities and the public sector in general should face the possible lack of AIrelated skills with synergy and identify new models for working with AI experts from the private sector and academic world. On the ia.italia.it portal, a mapping of AI-related courses in Italy will soon be available. In the meantime, the evolution of the Italian ecosystem can be observed (https://ia.italia.it/ia-in-italia/) related to start-ups, research centers and other realities that deal with the production or use of AI solutions.

Stay updated

The reference portal of the Artificial Intelligence Task Force, ai.italia.it will remain active and updated with the initiatives that will be deployed in the coming months. It is advisable to consult it frequently and to subscribe to the newsletter to stay updated on developments and news. The community, will also remain active, where it is possible to participate in open (or start new) discussions on issues related to the Task Force activity and to AI in general. For any information or any collaboration, there is always the email intelligenza-artificiale@agid.gov.it