Introduction
In November 2024, the government of the Generalitat constituted the CETRA (Commission of Experts for the Transformation and Reform of the Administration) with 12 leaders in charge of reviewing how to reform different administrative areas , each coordinating a team of 5 experts and coordinating with the other leaders. In the act of constituting this enormous team of almost 70 advisors, the presidential advisor asked that these experts not limit themselves to diagnosing, but rather to identify actions that could be implemented very quickly and we set ourselves the ambitious goal of having a third of the measures implemented by November 2025 .
I had the pleasure of being called to lead the team for the comprehensive transformation of the administration (CETRA-TDIA Group). The mandate was to propose what role Artificial Intelligence should play in the transformation of the digital administration of the Generalitat and I coordinated with Juan de la Paz , the leader of the group for transparency and traceability of procedures .
For our task, I chose a team of experts with great knowledge, complementary specializations and a solid professional career:
- Karina Gibert, leader at CETRA-TDIA
- Pau Fonseca, Associate professor and researcher at the Inlab (UPC)
- Laura Sunyol,Director of Innovation and Digital Transformation of the Mataró City Council
- Belén Arribas, Lawyer. President of ENATIC
- Encarna Pau, Responsible for cybersecurity of operational technologies at EMEA level. Marelli
- Karma Peiró,Executive Director of Visualization Transparency Foundation (VIT)
- Alícia Coca, Head of Business Technologies. Ferrocarrils de la Generalitat de Catalunya
We began by analyzing the scope of action and were able to verify that since Artificial Intelligence is transversal, the entire organization could be impacted by the adoption of this technology with different purposes and objectives. We started from a basic purpose “ That the adoption of AI within the Catalan public administration is articulated from a safe, ethical and reliable deployment of AI ” and we agreed on the vision of proposing a reform that moves towards the Society 5.0 of the future, what we call Administration 5.0 , and that uses all the technology at hand, including artificial intelligence to put the person at the center and make public administration a service for citizens.
Proposals
CETRA-TDIA group has ended up being structured into 5 major proposals that were delivered in a compendium with those of the other 11 CETRA groups to President Illa in a solemn ceremony on November 3rd and can be consulted on the CETRA website https://presidencia.bibliotecadig i tal.gencat.cat/handle/20.500.14623/947
CETRA-TDIA’s proposals focus on the introduction of AI in digital administration and are coordinated with other proposals from other groups on clear communication, transparency and traceability, etc.
1 Deploy an Artificial Intelligence governance model and promote it in the administration
Administration 5.0 is sought , aligned with Society 5.0 and which therefore puts citizens and people at the center. Doing so with guarantees requires an advanced AI governance model. CETRA- TDIA has identified the elements of the governance model and a proposal is made that starts from the general purpose , identifies expected benefits and values that determine the approach of the proposal. The proposal starts from the essential purpose that the adoption of AI within the Catalan public administration is articulated from a safe, ethical and reliable deployment of AI . and rests on the values proposed by the European Commission, and places special emphasis on values of transparency and traceability, responsible use of AI, justice, privacy, security . AI governance necessarily requires a good data governance model aligned with these values and a centralized data model with a federated data architecture is proposed, which reduces latency and develops a metadata layer rich enough to later allow successful exploitation of the data with data science processes. The proposal considers only a use of AI without agency, focused on assisting public professionals (and eventually citizens), but not oriented towards making automatic decisions or replacing professionals. People must continue to hold responsibility for decisions and focus on explainable AI models and solutions that argue predictions and recommendations with transparency and traceability without exceptions.
The model is articulated around three broad dimensions: instruments, processes and actors involved. The model is universal in scope, in the sense that AI represents a transversal digital infrastructure that requires a profound cultural change and affects the entire organization, all processes, all actors and the relationship also with the organizational ecosystem and people who interact with the administration from any role and position, and with any objective. Special emphasis is placed on the need to adapt cybersecurity levels to the potential vulnerabilities of shared data and cloud technologies. A special space is also dedicated to accountability mechanisms and, above all, to informing citizens about the use of AI algorithms.
2 Incorporate specialized AI talent and create a General Directorate of AI and Data
The adoption of AI as a basic digital infrastructure of the public administration will permeate the Generalitat with AI solutions and processes that use it. Software will need to be contracted and projects developed that require public employees with knowledge and skills in AI. It is essential that the Generalitat has specialized knowledge in AI internally to ensure that decisions on internal uses, ( sub )contracting, acquisitions of tools, developed projects, etc., are made with the greatest solvency and the lowest risks. CETRA-TDIA ‘s proposal is to establish a functional area specialized in Artificial Intelligence to support the entire administration in tasks related to AI. This functional area is best organized under its own general directorate that reports directly to the secretariat of telecommunications and digital transformation and focuses specifically on transformation using the data, digital transformation and AI as driving forces . Given the intimate link between AI, data, process efficiency and, in general, digital transformation, the proposal envisages that this general management covers these transformation objectives. CETRA-TDIA proposed for the general management a Professional profile with proven experience from a technical point of view in Artificial Intelligence, Ethical sensitivity and aligned with the principles of an AI for the common good, Aligned with the principles of a public service focused on people, With good knowledge of the internal functioning of the administration, With a vision of the country and with good leadership and management skills of collaborative teams. The position was called in the spring and the position was appointed last October. Regarding the specialized body On the other hand, the lack of talent in these areas requires deploying a talent acquisition program under competitive conditions in the market, if we want the most qualified people to be part of the administration. For this reason, a special Catalunya Futur program (also proposed by CETRA) is proposed to attract talent specialized in AI. CETRA-TDIA has developed the recommended professional profiles for this functional area, and the draft of the Catalunya Futur-AI program.
3 Integrate the use of Digital Twins in the innovation initiatives of the Generalitat
The proposal consists of deploying a new innovation methodology that we call TOBDI. (Transformation Oriented by Intelligent Digital Twin) that changed the approach to innovation throughout the administration, using advanced Artificial Intelligence tools, which go beyond generative AI and chatbots to agilely build small digital twins (models that simulate a real context digitally for prior testing and analysis) as the first step of any innovation process. This is a profound cultural change that would allow to accelerate innovative processes in the objective of the transformation towards an Administration 5.0. It will be important to establish very well how all digital twins should be integrated in a modular way into a single large digital twin that represents the entire administration and even society that can become in the medium term the basic tool for informed decision-making for the development of public policies , but also for continuous improvement of the administration itself. A methodology is provided to design small twins from flow charts that describe the processes that are to be improved and automatically translate them into executable language by a digital twin platform. An example case is provided on a possible modification of digital signature processes and it is illustrated how twins are derived from flow charts and by simulating with the twins the impact of the in silico innovation can be evaluated , which allows more refined proposals to be brought to a real pilot with a higher success rate. The savings in personnel hours and money that the proposal that improves is evaluated at any time are also quantified from the twin
As part of this proposal, the CETRA-TDIA group has worked to identify a set of main bottlenecks in the current functioning of the administration, and it is proposed to incorporate the TOBDI methodology to analyze how AI can improve these processes and services, prioritized according to their level of impact for citizens and their internal complexity. The individualized sectors are Intervention, Legal, Subsidies, Contracting and Notifications.
.
4 Deploy the intensive dissemination of recommendations for the good use of Artificial Intelligence in the Administration
The emergence of generative AI is spreading very quickly and favors the reduction of digital barriers, since it offers the possibility of interacting in natural language and does not require specialized profiles for its use, although there are risks of improper use.
Well, in this proposal, some impact decalogues have been created in the form of easy-to-understand infographics for the ethical and responsible use of generative AI aimed at end users and directors of AI units, as well as an extensive document that details the 10 principles of the decalogues and is linked to the infographics through a QR.
The group proposes using these infographics-decales as a basis for designing massive and periodic information campaigns to raise awareness and train the greatest possible number of public employees in the proper use of AI, with the idea of promoting the vision of “if you use it, do it well.”
However, material is also provided in the form of a decalogue for leaders of AI users in different departments with a practical collection of 10 principles to take into account when managing teams that use generative AI.
Campaigns are proposed to disseminate these Decalogues massively throughout the administration of the Generalitat, both physically and digitally, encouraging them to reach the largest possible set of public servants. The Decalogue with the recommendations for end users and the dissemination measure can be consulted at https://enginyeriainformatica.cat/dones-coeinf/decaleg-us-responsable-ia-generativa-empleats-publics/
5 Establish software quality standard levels applicable in the field of AI and supplier relations
Finally, the quality of a software determines its usefulness and suitability, its use with guarantees, and also its life cycle, that is, for how long it will be able to function correctly. The quality of the software has a direct impact on citizens when we talk about public sector software, as it determines the quality of public services, and also the return on the public investment made in the development/acquisition of that software.
CETRA-TDIA recommends establishing the mechanisms and procedures to have a standard of the Generalitat that sets out how to define, specify, structure, implement, validate and deploy software, including those that use artificial intelligence to guarantee not only its quality, extend its useful life, facilitate integration with other parts of the administration, etc., but also break a dynamic that, due to the structure of the Catalan economy, has not helped much so far. If there is a sufficiently complete standard, it is no longer essential who develops it and this opens the door for small entrepreneurs, or start-ups ups with highly technical and disruptive knowledge can contract parts of large orders directly with the administration. Breaking the rules and making those who have the knowledge develop regardless of whether or not they can later do the maintenance, and thus increasing the quality of the software generated and also the necessary fragmentation of the market to allow dynamization of a fabric like that of this country with a lot of entrepreneurship and SMEs capable of providing great disruptive value to the administration. It is key that the development of AI follows the highest quality standards that guarantee, among other things, robustness and lack of bias.
In this own standard, CETRA-TDIA recommends giving precedence to co-creation methodologies with all the actors involved in a software application (including citizens) (and avoiding situations like the recent verifactu affair). https://new.enginyeriainformatica.cat/news/comunicat-verifactu/ ) and user experience assessments before moving the solutions to production. The standards developed within the Generalitat must assume, among others, the criteria for traceability of procedures and the clear communication guide proposed by the other CETRA groups
Final considerations
Of the 5 measures proposed by CETRA-TDIA, 2 were implemented as of November 3 and the other 3 are currently in the hands of the AI commission of the Generalitat, developing the plan for their deployment to address the bases of a transformation as disruptive as it is pioneering that must allow public administration to evolve to a new model where service to citizens is transparent , the digital gap between public servants and citizens disappears and efficiency represents the bureaucratic standard of the future.
