Watch:
A recent paper published in Technovation marks an important step in the evolution of research on innovation districts by introducing new AI-based methods for predicting the innovation performance of city neighbourhoods.
The study builds on a long tradition of research on the geography of innovation, which began in the 1980swith the revival of industrial district theory by Giacomo Becattini through research on industrial districts of Italy. This research demonstrated how geographical concentration of specialised firms, local labour markets, and flexible supply chains can generate innovation. The perspective was subsequently broadened by Allen Scott and Michael Storper, who developed theories of industrial agglomeration and regional production systems, explaining how innovation is shaped by spatial organisation, agglomeration economies, and untraded interdependencies among economic actors. Over the following decades, this line of research evolved through concepts such as technology poles, high-technology clusters, regional innovation systems, learning regions, and innovation districts.
The new Technovation paper takes this literature a step further. Rather than explaining why innovation emerges in particular urban locations, it asks whether the innovation performance of city districts can be predicted before it becomes visible through conventional indicators such as patents, start-ups, or R&D activity.
Using artificial intelligence and machine learning, the paper analyses a rich set of neighbourhood characteristics, including business activities, urban functions, socio-economic conditions, and geospatial information, to identify the factors that best predict innovation performance. This methodology demonstrates that innovation potential can be inferred from combinations of urban characteristics that are difficult to detect using traditional statistical methods. The paper represents a significant methodological advance. The contribution is not a new theory of innovation districts but rather a predictive framework that supports evidence-based urban innovation policy. Instead of relying only on historical innovation indicators, planners and policymakers can use AI-based models to identify neighbourhoods with latent innovation potential, prioritise investments, and design interventions that strengthen emerging innovation districts. This methodology illustrates how artificial intelligence is transforming urban innovation research. AI is not used to analyse innovation after it has occurred; it can reveal the urban conditions that precede innovation and improve our ability to anticipate where new innovation is likely to emerge.
For URENIO, this work is particularly relevant because it extends the long-standing research agenda on innovation spaces, the geography of innovation, and the relationship between intelligence and innovation towards predictive urban analytics. It highlights how advances in AI and urban data science are opening new opportunities to understand, measure, and support innovation at the scale of city districts, where most of today’s urban transformations takes place.
The PDF is available here