Recommendations

Here is a list of recommendations, drawn up taking into account the 9 challenges presented in this document and the suggestions received through the public consultation. These recommendations are not to be considered definitive or immutable, but they evolve with the continuous public debate on Artificial Intelligence.

  1. Promote a national platform dedicated to the development of AI solutions in order to:
  1. promote the collection of annotated data, codes and learning models;
  2. organize and convey tests openly before the release of AI systems used in the PA in order to evaluate their behaviour and limit the anomalies and the amplification of the bias;
  3. offer adequate computing resources to experiment with techniques and solutions;
  1. Disclose to the public the intermediate results of the elaboration of AI algorithms (ex. parameters of neural networks) operated on data from public administrations, subject to conditions that may harm the privacy and security of citizens. These results must allow the reproducibility of the processes, their evaluation and verifiability;
  2. Enable with new resources the computational linguistic systems for the Italian language (such as digitized lexicons or annotated corpora) to be distributed with open licenses, in order to favour the development of services based on the treatment of natural language;
  3. Develop adaptive customization and recommendation systems that facilitate interaction with the services offered by public administrations based on the specific requirements, needs and characteristics of citizens. These systems can also be used to identify critical issues that hinder the improvement of public services;
  4. Promote the creation of a National Competence Centre that is a point of reference for the implementation of AI in the PA and that can provide predictions of impact and measurement of the social and economic effects of AI systems, in order to enhance the positive effects and reduce risks. The centre of competence must also propose a Manifesto with the cardinal principles of the implementation of AI in the PA;
  5. Facilitate the dissemination of skills through the promotion of certification of professionals working in the area of AI (through the creation and adoption of a shared framework ) and provide for the establishment of training paths for the inclusion of workers with ability to understand and implement AI solutions in the public administration (for example through specific courses at the National School of Administration). The numerous professional skills lacking are an opportunity to think about training courses focus on lasting and sustainable gender equality, both from a numerical point of view (ex. STEM graduates) and from a financial point of view (ex. remuneration equality);
  6. Provide a PA 4.0 plan to encourage PA investments in AI solutions starting from a call for challenges through open innovation tools and procurement for innovation;
  7. Support the collaboration between research, business accelerators and innovation hubs, both public and private, also at European level, to promote the adoption of AI solutions in the public sector;
  8. Establish a Trans-disciplinary Centre on AI, in synergy with the Skills Centre, which will have the following tasks:
    1. promote and disclose the debate on the evolution of ethics;
    2. support critical reflection on emerging ethical issues;
    3. improve the conditions for involving experts and citizens to transform technical and social considerations into regulations, standards and technical solutions;
  9. Define guidelines and processes based on the principle of security-by-design in the use of AI, increasing the levels of control and facilitating the sharing of data on cyber attacks to and from AI by all European countries.