Problem

The paper addresses a significant gap in the literature concerning the joint evolution of Artificial Intelligence (AI) and robotics. While previous research has examined AI and robotics separately using either adoption data for robots or patent indicators for AI, there has been a lack of systematic study distinguishing between traditional rule-based robots and those embedding AI functionalities. Furthermore, the influence of country-specific innovation systems on these technological trajectories has remained largely unexplored. The core problem is to empirically differentiate these domains and understand their long-run dynamics across various institutional environments.

Method

The authors constructed a novel patent-family dataset from PATSTAT spanning 1980–2019. They developed a unique methodology to define three mutually exclusive technological domains: core AI, traditional robots, and AI-enhanced robots. This involved combining Cooperative Patent Classification (CPC) codes, keyword-based identification in titles and abstracts, and document-level text mining using Python. To analyze the long-run dynamics, they employed time-series econometric methods, including tests for non-stationarity (Augmented Dickey-Fuller, KPSS), identification of structural breaks (Bai and Perron methodology), and pairwise cointegration analysis. This framework allowed them to assess common stochastic trends and co-evolution patterns across different technological domains and countries.

Results

Three main findings emerged from the analysis. First, patenting activity in core AI, traditional robots, and AI-enhanced robots followed distinct long-run trajectories: core AI showed rapid but uneven growth, traditional robots exhibited gradual growth, and AI-enhanced robots experienced a sharp acceleration from the early 2010s. Second, structural breaks predominantly occurred after 2010, particularly for core AI and AI-enhanced robots, indicating a common shift in innovation dynamics linked to AI diffusion. Traditional robots, however, displayed earlier breaks. Third, long-run relationships between these domains varied systematically across countries. China exhibited strong cointegration between core AI and AI-enhanced robots, while the United States showed weaker integration. Europe, Japan, and South Korea presented intermediate and technologically differentiated patterns.

Implications

The findings suggest that the relationship between AI and robotics is a process of differentiated technological co-evolution, not a uniform diffusion pattern. The economic impact of AI as a general-purpose technology depends not only on its expansion as an inventive field but also crucially on the institutional and sectoral contexts in which it is embodied. This research provides a more precise understanding of how pervasive technologies propagate through applied technological systems. Future work could investigate the determinants of cross-country differences, the firm-level and labor-market effects of AI-enhanced robots, and patent quality/influence, and extend the dataset beyond 2019.