![Sample our Social Sciences journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110822/TOC-Subject-Sample-2021-Social-Sciences.jpg"})
Abstract
For assessing the overall impact of Artificial Intelligence (AI), it is crucial to continuously monitor large corporations. This paper delves into the examination of 42 corporations that rank among the world's largest investors in R&D, accounting for over one-third of AI patents globally. The focus is on their post-patenting performance, specifically in terms of employment changes, and comparing it with the outcomes of 42 similar companies operating in the same sectors. The latter also recorded substantial levels of R&D expenditures but were not significantly involved in AI patenting. The key findings reveal that, in the medium – and high-tech manufacturing sectors, companies with the highest proportions of AI patents incurred in employment reductions. Conversely, IT services companies experienced substantial employment growth. Along with tentative explanations of these findings, advantages, limitations, and possible developments of this type of analysis are illustrated in the concluding section.
Acknowledgments
I thank three anonymous reviewers for their comments and suggestions.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 In reviewing the following studies, the procedures they have adopted for collecting AI patents are neglected. All the examined studies employ specific and shared patent classification codes, and in most cases, the outcomes of keyword searches.
2 They also show that the labour-friendly effect of AI patents is larger and more significant for the companies established after 1990.
3 Since many companies included in the Scoreboard consist of diversified conglomerates, the imputation of a unique sector of activity represents a clear limitation of the database and, a fortiori, of the present study.
4 For instance, the International Patent Classification contains the specific code B25J9/16 – termed “Programme controls” – which includes inventions that very often connect robots to AI systems emulating the decision-making ability of a human expert.
5 Among the excluded companies there are Amazon and Alibaba since it was impossible to compare them with similar companies with high R&D expenditures. This also applies to Boeing and Honeywell (mainly active in aerospace and defence systems), General Electric (aviation, healthcare, and energy), and Philips (lighting and personal care). The latter two can be hardly classified as Electronic & Electrical Equipment companies and included in the broad sector of Computers & electronics (see and Table A.1 in Appendix). Finally, for the Japanese Fanuc (automation products and services) and the Chinese Leshi (IT services) data on economic performances after patenting were incomplete.
6 The threshold for inclusion in the top 50 patentees, in terms of number of AI patents, cannot be determined because the dataset of the JRC-OECD project does not provide the total number of patents held by the world’s top investors in R&D.
7 Article 67(1) of the European Patent Convention states that from the date of its publication a European patent application provisionally confers on the applicant the same rights as would be conferred by a patent granted. Similarly, in the US, the American Inventor's Protection Act of 1999 states that the owner of a published application receives provisional rights to pursue royalties or infringement damages for the period between the date of publication and that of patent grant (cf. Hegde, Herkenhoff, and Zhu Citation2022).
8 The performance of Qualcomm (characterized by an increase of employment with a negative change of sales and high rates of profitability) seems quite at odds with expectations. The situation is even more contradictory if the analysis starts from 2016. For this reason, the rate of changes of employment and sales are computed from 2017 to 2019. In any case, due to its relatively low size in terms of employment, the exclusion of Qualcomm would not change our findings.
9 This inconsistency, which applies to Baidu too, would be much more evident by taking 2016 as the starting year. For this reason, the employment and sales changes of Hewlett Packard Enterprise and Baidu refer to the period 2017-2019.
10 “AI has extended the frontier of technological possibility towards boundaries that are barely visible at present. The tasks that machines will be able to accomplish, the rate at which new innovations may emerge, and the speed with which socially impactful technological innovations may diffuse is unknown.” (Autor Citation2022, 28).
Additional information
Notes on contributors
Alessandro Sterlacchini
Alessandro Sterlacchini is Full Professor of Applied Economics at the Università Politecnica delle Marche (Ancona, Italy) where he teaches Microeconomics (under-graduate level) and Economics of Innovation (post-graduate). His research activity has been mainly developed in the field of the Economics of Innovation, with special focus on the determinants and economic effects of R&D, patents, and innovations. In this connection, he carried out and published in international refereed journals several empirical studies concerned with industries, firms, and geographical areas.