When talking about artificial intelligence, it is common to think of algorithms, automation, virtual assistants, predictions or tools capable of generating texts, classifying documents or helping to make decisions. But in a public administration, and especially in a city council, the first question should not be "what artificial intelligence tool can we buy or try?".
The first question should be something else, such as:
- What public problem do we want to solve?
- Do we want to reduce processing time?
- Do we want to improve citizen care?
- Do we want to detect social needs before they become more serious?
- Do we want to better organize incidents on public roads?
- We want to facilitate the transparència?
- Do we want to help municipal staff classify documents or files?
- Do we want to make better use of the data we already have?
Artificial intelligence can help with many of these challenges. But it will only do so well if the starting data is reliable, well described, protected, and can be used with guarantees. That's why, before talking about algorithms, we need to talk about data.
Let's continue on the path: inventory, metadata, quality, master data and roles
In the previous posts in this series we have been building a background idea:
We have explained What does it mean to govern data in a city council?, Why do we need to know what data we have?, how metadata helps us understand them, why incorrect data can end up generating bad service, why master data (people, territory, facilities, entities or suppliers) are so important and what roles and responsibilities need to start identifying.
This journey does not precede artificial intelligence by chance. It is its basis..
- A city council that does not know what data it has will hardly be able to use it to train, feed or validate an AI system.
- A city council that does not describe its data with metadata will have difficulty knowing where it comes from, what it means, or if it can be reused.
- A city council that does not review the quality of data may obtain erroneous results.
- A city council that does not have clear master data can feed AI tools with duplicate or inconsistent information.
- A city council that does not know who does what will have difficulty validating, supervising or assuming responsibility for the results.
AI does not replace data governance: it needs it.
Good data before good algorithms
An algorithm can be very advanced, but if it works with incomplete, erroneous, outdated or biased data, the results will also be.
This is especially important in public administration, where decisions can affect rights, services, resources, priorities and relations with citizens.
For example:
- If a system classifies requests but historical files are mislabeled, it will learn from previous mistakes.
- If a tool analyzes social needs but the data is incomplete or does not well represent certain groups, it can generate unfair conclusions.
- If an automated assistant provides information to citizens but the knowledge base is not updated, it may give incorrect answers.
- If a tool transparència anonymizes documents but there are no clear criteria regarding personal and sensitive data, it can generate risks.
Therefore, the quality, context, security and oversight of the data are as important as the artificial intelligence model used.
Public AI must have purpose
Not every application of artificial intelligence makes sense for a city council. AI should not be a fad or a race to incorporate technology. It should respond to real needs and generate public value.
A small municipality may not need to prioritize large predictive systems, but tools that help reduce administrative burdens, classify documents, guide inquiries, prepare drafts, detect errors, summarize information or facilitate the work of municipal staff.
The key question is: what specific benefit does it bring to the citizenry or the organization?
Purposeful AI should help:
- Provide more agile services
- Reduce repetitive tasks
- Improve the quality of information
- Facilitate access to services
- Strengthen the transparència
- Support decision-making
- Improve planning
- Detect inequalities or needs
- Free up municipal staff time for more valuable tasks
But this purpose must be well defined before we begin. Technology should not seek out a problem. The public problem must guide the technology.
Ethical, safe and guaranteed AI
The use of artificial intelligence in the public sector requires specific guarantees. It is not enough for a tool to work. It is necessary to know how it is used, with what data, for what purposes, with what limits, with what supervision and with what risks.
A public AI must be understandable, proportional, safe, overseeable, transparent, explainable and respectful of people's rights.
This involves asking yourself questions like:
- What data does it use?
- Does it contain personal or sensitive data?
- Is the purpose legitimate and clear?
- Is there a risk of bias?
- Who validates the results?
- Can you explain how a recommendation was reached?
- Is there a person responsible for the final decision?
- Can the operation be audited?
- Can errors be corrected?
- Have the affected citizens or personnel been adequately informed?
These questions are not obstacles to innovation. They are conditions for responsible public innovation.
Trust is essential. And without trust, public AI will not be sustainable.
Digital sovereignty also involves data and AI
When a city council uses data and artificial intelligence, it must also think about its control capacity.
- Where is the data?
- Who can access it?
- Can they be exported?
- Can they be reused in other services?
- Does the provider completely condition the use of the information?
- Can the solution be changed without losing knowledge?
- Is the model used transparent?
- Is municipal data used to train third-party systems without sufficient control?
These issues are part of digital sovereignty. For municipalities, this is especially important. They often have less capacity for individual negotiation with suppliers and fewer resources to evaluate complex solutions. For this, shared criteria, common services and support are needed.
Digital sovereignty doesn't mean doing everything alone. It means being able to make informed decisions, avoiding unnecessary dependencies, and ensuring that public data is managed in the service of the general interest.
Territorial equity: AI should not be just for large municipalities
Artificial intelligence can generate opportunities, but it can also widen inequalities if only the most resourced entities can take advantage of it. One of the great challenges in the local world is to ensure that small and medium-sized municipalities can also access knowledge, tools, services and support. Local public AI must be thought of from a territorial equity perspective.
This means that solutions must be proportionate, reusable, shared and adaptable to different municipal realities.
Not all municipalities have the same size, the same data, the same systems or the same equipment. But all should be able to benefit from a smarter, more agile and more citizen-oriented administration. Equality of digital opportunities must also be guaranteed between administrations.
Local.iA: A framework for deploying public, ethical, sovereign and purposeful AI
In this context, Local.iA, promoted by the Barcelona Provincial Council, is becoming a benchmark initiative to lay the foundations for the deployment of artificial intelligence in the local world with public criteria.
Local.iA proposes a conceptual and strategic framework to help local administrations deploy AI in a responsible and ethical manner, with instruments, models, training resources and assistance.
This approach is particularly relevant because it does not treat AI just as a technology, but as a public capability that must be deployed in a conscious, democratic, transparent, ethical, sustainable and safe manner.
It also incorporates a view of territorial equity; the deployment of AI cannot be uniform, but must take into account the diverse reality of local entities and their capacities. It fits perfectly with the need for small and medium-sized municipalities not to be left behind.
A shared strategy with the AOC and the Smart Local Government Network
Local.iA is part of a broader framework of institutional cooperation. The Institutional Agreement for the Promotion of Smart Local Governments aims to modernize and improve public services through the use of data and artificial intelligence, ensuring an ethical and responsible approach. This agreement was signed between the Generalitat de Catalunya, Localret, Barcelona City Council, the four Catalan provincial councils and the AOC Consortium.
This shared strategy is essential so that the deployment of AI in the local world is not fragmented into isolated initiatives, but rather advances on common grounds: data governance, shared services, ethical criteria, digital sovereignty, training and reuse of solutions.
In this framework, the AOC Consortium plays a key role through the Local World Data Space and common services aimed at enabling local entities to take advantage of data and AI without having to start from scratch.
Currently, work is also underway to extend Local.iA with the other Catalan provincial governments. This process should allow progress towards a vision of the country, where the experiences, learning and resources generated can benefit all local entities in Catalonia.
Use cases: start with specific problems
For small and medium-sized municipalities, the best path is not to start with large AI projects, but with specific and well-defined use cases.
Some examples could be:
- automatically classify instances or requests;
- help anonymize documents before publishing them;
- detect duplicates or errors in municipal data;
- summarize files or supporting documentation;
- improve internal information search engines;
- support citizen care with updated knowledge bases;
- identify patterns in road incidents;
- help prepare management indicators;
- facilitate consultation of internal regulations or procedures;
- detect incomplete information before processing.
These cases can have a lot of value if they are well defined. But in all of them, well-organized data, quality criteria, metadata, responsibilities, human oversight and legal guarantees are needed. AI can help, but it cannot replace public accountability.
Human oversight: public decision-making is not delegated to a machine
A basic principle in the use of AI in the public sector is that responsibility remains human and institutional. A tool can recommend, classify, summarize or detect patterns. But the city council must retain the ability to validate, correct, review and decide. This is especially important when AI can affect rights, access to services, prioritization of resources or processing of personal data.
Human supervision is not a procedure. It is a guarantee. It is also a way to ensure that the professional knowledge of municipal staff remains central. AI should help people working in municipalities, not replace their public judgment.
From data culture to responsible AI culture
Data culture and artificial intelligence culture are not two separate things, but very related. An administration that does not have a data culture will hardly be able to have a responsible AI culture. AI requires understanding what data is, how it is generated, who maintains it, what quality it has, what risks it incorporates and how it can be used legitimately and usefully. For this reason, the previous posts are also a preparation for AI.
Inventorying data, describing metadata, improving quality, organizing master data, and defining roles are not prior tasks without visible value. They are the foundations that will make possible a safer and more effective use of artificial intelligence.
An AI at the service of the local world
Artificial intelligence can be a great opportunity for local authorities. It can help make administrations more agile, more efficient, more proactive and closer to the people.eres. It can free up time, improve attention, strengthen transparència and help make better decisions.
But it will only generate public value if it is deployed on good data, with good governance, with ethical criteria, with digital sovereignty, with human supervision and with territorial equity.
Small and medium-sized municipalities should not see AI as a distant or inaccessible technology. They should see it as an opportunity that must be built together.
Local.iA, the Smart Local Government Network, the Institutional Agreement and the Local World Data Space offer a basis for taking this path together. Because local public AI must not only be smart. It must be useful, reliable, ethical, sovereign and aimed at improving people's lives.
