AI-driven recommendations are reshaping cities, Italian study finds

Artificial intelligence is not only changing how people move around cities, but also how cities themselves evolve. A new Italian study coordinated by researchers in Pisa shows that the recommendations generated by popular digital platforms can gradually reshape urban flows, concentrating activity in a limited number of places while leaving others increasingly marginal.

The research was carried out by the National Research Council through its Institute of Information Science and Technologies in Pisa, in collaboration with the CNR Institute for High Performance Computing and Networking in Palermo and the Scuola Normale Superiore. The study has been published in the scientific journal Machine Learning.

Algorithms as “urban actors”

The researchers analysed so-called recommender systems, the algorithms behind apps such as Google Maps, Tripadvisor, Yelp and TheFork, which suggest places to visit based on users’ past behaviour and aggregated data.

According to the study, this is the first time that the feedback loop between algorithms, human behaviour and urban space has been explicitly modelled. Instead of evaluating algorithms in isolation, the research treats them as active participants in city life, capable of influencing how people move, where they gather and which areas gain or lose visibility.

Individual variety, collective concentration

At an individual level, algorithmic recommendations can encourage people to explore a wider range of places. Over time, however, the collective effect tends to be the opposite. Traffic, visits and social activity increasingly concentrate in a small number of already popular locations, while less visible areas struggle to attract attention.

To study these dynamics, the team developed a simulator that reproduces the interaction between humans and AI in an urban context. The model makes it possible to observe not only individual choices, but also broader trends such as the polarisation of neighbourhoods and the growing imbalance between central and peripheral areas.

Implications for cities like Florence

Although the study does not focus on a single city, its findings are particularly relevant for historic and tourist-heavy urban centres such as Florence, where digital recommendations strongly influence visitor behaviour. Concentration effects can amplify pressure on already crowded areas, while reinforcing economic and social gaps within the urban fabric.

The researchers argue that greater civic awareness is needed when designing and regulating urban-facing algorithms. In their view, future recommender systems should not optimise only individual convenience, but also consider spatial equity, accessibility and the overall social health of cities.

❤️ Support Florence Daily News

If you liked this article, please consider supporting Florence Daily News.

We are an independent news site, free from paywalls and intrusive ads, committed to providing clear and reliable reporting on Florence and Tuscany for everyone.

Your support — whether a one-time gift or a regular contribution — helps us stay independent and keep telling the stories that matter.

Donate securely via Stripe below.