Skip to content

Olivier Decrock Deputy Mayor of Saint-Ouen-sur-Seine

decrock.fr

Understanding the issues

Artificial intelligence: understand, govern, act

AI is already transforming many sectors. At city level, it can help organise certain services better, make information more accessible or support innovation. But it also raises major questions: data protection, bias, transparency, human oversight and citizen trust.

Artificial intelligence should not be thought of as a technology that decides in place of humans, but as a tool that can help to better understand, better organise and better serve, under human oversight, with transparency and responsibility.

What are we talking about?

Artificial intelligence refers to a set of technologies capable of analysing information, identifying patterns, generating content or helping to formulate answers. It is already present in many everyday uses: machine translation, search engines, recommendations, digital assistants and writing aids.

But AI is not human intelligence. It does not understand the world like a person, has no moral judgement and can produce errors. This is why its use in public services must be governed. Useful AI is AI whose purpose, limits, data and conditions of human oversight are known.

What AI can change for a city

At city level, artificial intelligence can be useful if it addresses concrete needs. It can help guide residents through their procedures, make certain content more accessible, translate or simplify information, support employees in summarising tasks, or analyse aggregated data to better understand the needs of the area.

However, AI must not become a black box. It must remain a support tool, under human oversight, with clear objectives and strong guarantees. In public services, efficiency is not enough: equal treatment, transparency, data protection and the ability to understand decisions must also be ensured.

The red lines of public service

No opaque automated decisions

A decision affecting a person must not be made by an incomprehensible or uncontrolled system.

No use without a clear purpose

An AI tool must address an identified and explicit need.

No excessive data collection

The data used must be necessary, proportionate and protected.

No replacing humans in sensitive situations

Social, administrative or individual situations must keep a human point of contact.

No unevaluated tools

Effects, errors, biases and impacts must be measured.

No digital exclusion

People who cannot or do not wish to use a digital tool must have alternatives.

AI and inclusion: avoiding new blind spots

Artificial intelligence can become a tool for inclusion when it helps translate, simplify, guide or make information more accessible. It can facilitate the understanding of complex procedures and support people facing linguistic, administrative or digital obstacles.

But AI can also create new blind spots. If it is designed without taking the diversity of situations into account, it can respond poorly to certain audiences, reproduce biases or reinforce inequalities. Inclusive AI must therefore be tested, explained, supported and always complemented by human solutions.

AI, employment and local skills

Artificial intelligence is already transforming the world of work. It changes certain tasks, creates new skills needs and requires everyone to develop a stronger digital culture. At local level, this transformation must be supported so as not to widen the gap between those who master these tools and those who endure them.

Training people in AI does not mean training everyone to become an engineer. It means providing reference points: understanding what a tool does, knowing how to check an answer, protecting one’s data, identifying risks, using AI in a useful way and keeping a critical mind. It is a matter of empowerment, employment and equal opportunity.

Generative AI: opportunities and vigilance

Generative AI refers to tools capable of producing text, images, summaries, code or other content from a user request. These tools can be useful for preparing a draft, rephrasing a document, translating information or making content more accessible.

But they must be used with caution. Generative AI can invent information, produce credible errors or reproduce biases present in its training data. It therefore replaces neither verification, nor editorial responsibility, nor human judgement. In public services, this vigilance is essential.

Should AI be feared?

Concerns about artificial intelligence are legitimate. They relate to employment, civil liberties, data protection, discrimination, surveillance and the loss of human contact. These questions must not be dismissed: they must be placed at the heart of any responsible approach.

AI should not be accepted because it is modern, nor rejected because it is new. It must be assessed against simple criteria: is it useful? Is it understandable? Is it controlled? Does it protect rights? Does it genuinely improve the service provided?

Frequently asked questions

Will AI replace public employees? +

The aim must not be to replace employees, but to help them with certain repetitive or time-consuming tasks, in order to free up time for reception, listening and support.

Can AI make mistakes? +

Yes. AI can produce errors, incomplete answers or biased results. That is why human oversight is essential.

Does AI use my personal data? +

It depends on the tools and uses. Every project must specify what data is used, why, with what guarantees and for how long.

Can AI be used without surveilling residents? +

Yes. Responsible use of AI must exclude any logic of generalised surveillance and strictly respect civil liberties.

Is AI always useful? +

No. Some needs are better addressed by administrative simplification, human support or better organisation. AI is not a universal answer.

How can discrimination be avoided? +

The data used must be analysed, the results tested, biases identified, limits documented and human oversight maintained.

A question, an idea, a suggestion?

Residents can submit a question, a difficulty, an idea or a suggestion related to the smart city, artificial intelligence or international relations.

Write via the contact form