ABSTRACT
Until now, the management literature on Artificial Intelligence (AI) focuses mostly on the diverse applications of this technology, while its development has attracted only limited attention. To partially fill this research gap, the present paper analyses a large sample of AI patents and investigates the potential determinants of their technological impact. We show how University-Industry (UI) collaborations seem to be less able to develop high-impact AI patents, compared to other types of partnership based on the involvement of either universities or companies. This result contrasts with the previous literature on the inventing process of other general-purpose technologies (GPT), thus clarifying how the development of AI may be significantly affected by the peculiar characteristics of this technology. Thereby, our findings not only shed further light on the inventing process of AI solutions but may also stimulate the debate on the development of other GPTs strongly imbued with scientific knowledge.
JEL:
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 These data are validated thanks to the publication of the complete list of AI patents by the UK Intellectual Patent Office, which allows us to minimise the possible errors in the replication of this study.
2 This result is further confirmed by an alternative regression model, not reported in the text, in which we tested how the technological impact of AI inventions is affected by the number of firm applicants and URI ones. Only this last variable has a significant and negative effect, thus further confirming the cons of having URIs among the applicants of AI patents.