Automation and related technologies: a mapping of the new knowledge base

Abstract

Using the entire population of USPTO patent applications published between 2002 and 2019, and leveraging on both patent classification and semantic analysis, this paper aims to map the current knowledge base centred on robotics and AI technologies. These technologies are investigated both as a whole and distinguishing core and related innovations, along a 4-level core-periphery architecture. Merging patent applications with the Orbis IP firm-level database allows us to put forward a twofold analysis based on industry of activity and geographic location. In a nutshell, results show that: (i) rather than representing a technological revolution, the new knowledge base is strictly linked to the previous technological paradigm; (ii) the new knowledge base is characterised by a considerable—but not impressively widespread—degree of pervasiveness; (iii) robotics and AI are strictly related, converging (particularly among the related technologies and in more recent times) and jointly shaping a new knowledge base that should be considered as a whole, rather than consisting of two separate GPTs; (iv) the US technological leadership turns out to be confirmed (although declining in relative terms in favour of Asian countries such as South Korea, China and, more recently, India).

Appendices

Appendix 1

In this technical appendix we formally define the two proximity measures, namely cosine similarity and Spearman rank correlation, used in the construction of Table 8 and discussed in Sect. 4.3. As extensions to the underlying core-periphery levels are straightforward, we only explain their development in the overall case.

Once a group of patents are matched to their corporate assignee(s) (cf. Sect. 3.3), it is possible to build a rank of their corresponding sectoral industries, sorted by frequency of occurrence. Provided that there exist 99 NAICS codes at the 3-digit level, the ranking can be expressed as a vector in the 99-dimensional vector space of natural numbers. Given two such vectors ${\rm X},{\rm Y}\in {\mathbb{N}}^{99}$ corresponding to, say, the whole sets of robotics and AI patents, respectively (or any of their core-periphery subsets), it is possible to define their cosine similarity as the cosine of the angle between them, which is also equal to the inner product of the same vectors normalised to unit length. Formally,

$${\text{cos}}\left( {X,Y} \right): = \frac{{X \cdot Y}}{{\left| {\left| X \right|} \right|\left| {\left| Y \right|} \right|}} = \frac{{\sum _{i} x_{i} y_{i} }}{{\sqrt {\sum _{i} x_{i}^{2} } \sqrt {\sum _{i} y_{i}^{2} } }}$$

where ${x}_{i}$ and ${y}_{i}$ denote the components of vectors $X$ and $Y$, respectively, and $||\cdot ||$ denotes the Euclidean norm. Since rank vectors are non-negative, values of their cosine similarity are bound to the unit interval $\left[{0,1}\right]$.

In a similar fashion, it is possible to define the Spearman rank correlation as the usual Pearson correlation coefficient between the rank vectors $X$ and $Y$. Formally,

$${r}_{s}:={\uprho }_{XY}=\frac{{\rm cov}\left(X,Y\right)}{{\upsigma }_{X}{\upsigma }_{Y}}$$

Once these similarity measures are defined, it is possible to check whether the core-periphery architecture devised in Sect. 3.2 displays a satisfactory degree of inner consistency. Ideally, given the defined hierarchy, adjacent levels should bear more mutual similarity than non-adjacent ones. Accordingly, level CP1 should be closer to level CP2 than to level CP3, and closer to level CP3 than to level CP4, and level CP2 should be closer to CP3 than to CP4. Table 9 reports the cross-level proximity measures, both in terms of cosine similarity and Spearman correlation, for both robotics and AI patents, corroborating our core-periphery structure by validating the aforementioned requirement.

Table 9 Cross-level cosine similarity and Spearman rank correlation within the core-periphery architecture for both robotics and AI patents

Appendix 2

See Tables 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, and 22.

Table 10 Sectoral relevance to robotics patents for each core-periphery level in the first subperiod (2001–2007)

Table 11 Sectoral relevance to robotics patents for each core-periphery level in the second subperiod (2008–2013)

Table 12 Sectoral relevance to robotics patents for each core-periphery level in the third subperiod (2014–2019)

Table 13 Sectoral relevance to AI patents for each core-periphery level in the first subperiod (2001–2007)

Table 14 Sectoral relevance to AI patents for each core-periphery level in the second subperiod (2008–2013)

Table 15 Sectoral relevance to AI patents for each core-periphery level in the third subperiod (2014–2019)

Table 16 Country relevance to robotics patents for each core-periphery level in the first subperiod (2001–2007)

Table 17 Country relevance to robotics patents for each core-periphery level in the second subperiod (2008–2013)

Table 18 Country relevance to robotics patents for each core-periphery level in the third subperiod (2014–2019)

Table 19 Country relevance to AI patents for each core-periphery level in the first subperiod (2001–2007)

Table 20 Country relevance to AI patents for each core-periphery level in the second subperiod (2008–2013)

Table 21 Country relevance to AI patents for each core-periphery level in the third subperiod (2014–2019)

Table 22 Cosine similarity and Spearman rank correlations between robotics and AI core-periphery levels in each of the three subperiods