Bayesian inference of spatial and temporal relations in AI patents for EU countries

Area of Science:

• Economics
• Computer Science
• Innovation Studies

Background:

• The rapid growth of Artificial Intelligence (AI) patents presents complex spatial and temporal dynamics within the European Union (EU). Understanding collaboration patterns and trend evolution is crucial for policy and research.
• Existing models may not fully capture the nuanced interactions between countries or the specific growth trajectories of AI patenting.

Purpose of the Study:

• To develop and apply quantitative models for analyzing spatial and temporal behaviors of AI patents in EU countries.
• To identify and measure inter-country collaboration in AI patenting.
• To model and forecast the trends and intensity of AI patent applications.

Main Methods:

• Spatial analysis using Poisson regression to model collaboration between country pairs based on common patents.
• Bayesian inference to estimate interaction strengths between EU countries and the rest of the world.
• Temporal analysis employing an inhomogeneous Poisson process combined with logistic curve growth for trend modeling.

Main Results:

• Poisson regression identified significant variations in cooperation levels between different country pairs.
• Bayesian analysis revealed a notable lack of cooperation in AI patenting for specific country pairings.
• The inhomogeneous Poisson process model accurately captured the temporal trend of AI patenting, indicating an upcoming slowdown.

Conclusions:

• The study provides quantitative insights into the spatial and temporal dimensions of AI patenting within the EU.
• Identified cooperation deficits highlight areas for potential policy intervention to foster greater international collaboration.
• The predicted slowdown in patenting intensity suggests a maturing phase in AI innovation within the analyzed regions.