Artificial intelligence (AI) has emerged as a comprehensive field transforming diverse global sectors, from industry to healthcare and education. This essay, published in April 2024, offers a detailed analysis of the current AI landscape, based on extensive research of academic papers and patent data collected from the Lens.org platform, covering the period from January 2019 to March 2024. The primary objective is to bridge the gap between academia and industry by examining recent trends in AI research and development, thereby providing guidance for policymakers, investors, and scientists.
In summary, our goal is to map the global dynamics of AI research and collaboration, thereby guiding future efforts in this rapidly evolving field. This research offers a detailed and insightful view of the current AI landscape, highlighting the importance of collaboration between academia and industry to address challenges and capitalize on opportunities in this dynamic field.
Analysis of Research Documents: In 2023 alone, 74.426 articles on AI were generated.
Since 1978, the total number of registered research articles is 1,262,789: 198,163 from 2019 to 2024, with 2023 being the most productive year at 74,426 articles. The analysis shows that Asia, especially China, stood out as the leader in research output, with significant contributions from institutions such as the Chinese Academy of Sciences and Tsinghua University. Wei Wang was identified as one of the most influential authors, with 292 published articles and high citations in both research and patents. The predominant research topics include Computer Vision, Machine Learning, Deep Learning, and Neural Networks, reflecting the interdisciplinary nature of AI, which spans fields such as biology, medicine, and engineering.
AI patents: a field led by the United States.
In terms of patents, a total of 316,883 AI-related patents were registered during the same period, with the United States leading in patent ownership with 180,085 registrations, representing 56% of the total. China follows in second place, with 63,503 patents.
In the eastern part of the world, three countries lead the ranking after China: South Korea with 8,736 patents, Japan with 715, and Taiwan with 707. In the western hemisphere, the United States and Canada lead with 180,085 and 509 patents respectively, followed by the United Kingdom with 244. Australia has 507 patents. Latin America is led by Mexico, with 81 patents.
Prominent companies such as Samsung Electronics in the East and IBM in the West led the patent filings, with prominent categories including Machine Learning, Learning Methods, Network Combinations, Backpropagation, Recurrent Networks, Probabilistic Graphical Models and Inference or Reasoning Models, digital and image data processing, health informatics, and image or video recognition.
What methodological approach was chosen?
It is important to note that this document provides an overview of the last 5 years.
Using Lens.org, a collaborative data platform for open innovation, an extensive collection of AI-related academic literature and patent metadata was analyzed. Lens is an open initiative from MIT (Massachusetts Institute of Technology), freely available for research purposes, allowing users to explore related work from both academia and industry. This methodological approach enabled the creation of citation graphs, impact metrics, and statistical analyses derived from global patent datasets, providing a comprehensive view of AI research output and impact worldwide.
We used the Lens platform to extract data on research articles and patents because it allows for refining the search by selecting "Field of Study," while the patent tool does not offer this option. This can introduce an error in the total number of patents when compared to the total number of research articles. However, this does not invalidate the reported conclusions and percentages. This study should be validated with equivalent searches on other knowledge platforms, such as Scopus, which is specific to research articles, and Espacenet, which is specific to patents and supported by the European Patent Office. Nevertheless, it is possible to fully validate the results of this study for patents using other tools such as Google Patents, Questel, or Patsnap, although the latter two are limited by their restricted access.
Discussion and Conclusions
The analysis revealed a significant gap between academic and industrial collaboration in terms of patents and scientific publications, highlighting the urgent need for closer collaboration to optimize the social impact of AI innovations.
We observed that the number of patents (316,883) is greater than the number of articles (198,163); only 3,136 patents cite research articles (1.58%), and 6,511 articles cite patents (3.28%). While it is essential to keep innovation on research agendas, opportunities to acquire resources could be lost if there is insufficient focus on product development.
Globally, Asian research output was dominant, while the United States led in patent ownership due to policy and economic factors. Wei Wang emerged as the leading author, with 292 cited articles in 2,808 research papers and 19 patents. Thematic clusters such as Computer Vision, Convolutional Neural Networks, Adversarial Systems, Facial Recognition, Explainable AI, and Machine Learning Decision Trees dominated the research landscape.
In summary, our analysis reveals that Asia, particularly China, is the leading producer of research articles in artificial intelligence, accounting for 37.30% of total publications, followed by North America, primarily the US (13.41%), and Europe (9%). The authors' institutional affiliations are concentrated in Asia, with the Chinese Academy of Sciences leading the list with 5,112 articles, while Harvard University stands out as the only North American institution among the top contributors, ranking seventh.
We recommend the development of policies that foster greater collaboration between the academic and industrial sectors to drive continuous innovation in AI and close the identified gaps.
To be continued…
The next steps for this research team are focused on conducting a network analysis to better understand the topography of this ecosystem (academia, industry, and government). We plan to identify the main actors, fields of interest, and strategic areas of government development related to AI. Furthermore, it is important to replicate this study by region to understand the evolution of AI in each and to direct resources toward global interests, unifying efforts.
Quo Vadis IA It is a research line of the Faculty of Engineering of the Universidad AustralThe goal of this interdisciplinary team is to contribute to the understanding of the state of the art of Artificial Intelligence by providing information on future scenarios so that public and private research organizations, universities, and companies can define long-term policies and strategies.
