As artificial intelligence (“AI”) has been attracting attention as an innovative growth engine that will drive future industries and societies, strengthening the research capacity for AI is becoming a bottom-line issue. Research capacity is the most crucial intangible activity for technological innovation. It is key to achieving the best performance in the upcoming technology market. Therefore, we, in this article, sought to develop an AI Research Index that could measure the research capacity for AI for universities around the world and draw implications.
We defined AI Research Index as the indexed value of AI research performance from 2016 to 2019. We leveraged Scholarly Output, Citation per Publication (“CPP”), and the Field-Weighted Citation Impact (“FWCI”) to measure research capacity for AI. We then reflected weights for variables. We first selected the world's top 500 universities based on the Scholarly Output for AI research and measured the AI Research Index for 500 universities in consideration of the quantity, quality, and weight of variables. After that, we selected the world's top 100 AI universities and analyzed the proportion by country.
The world's top 500 universities were first selected based on their Scholarly Output for AI research. They were found to conduct an average of 404 research studies each over a four-year period (from 2016 to 2019). Nationalities of these top 500 universities are: 101 (20.2%) in China, 61 (12.2%) in the US, 45 (9.0%) in India, 29 (5.8%) in the UK, 25 (5.0%) in Japan, and 21 (4.2%) in France. In consideration of the quantity and quality of performance indices, we measured the AI Research Index and found that the average index was 46.01 for the top 500 universities. The average index of AI Research Index of the top 100 universities was 67.26, different from that of 500 universities. Among the top 100 universities, ratios of universities in China, the US, and the UK were high. Among these top 10 universities with the highest AI research indices, the US had the highest proportion at 40%.
The implication of this research was that there was a difference in AI research capacity between universities. The distribution of AI research capacity was not a normal one, but a form of power law. This result was similar to results of previous research studies on the distribution of human resources. The top universities with high AI research indices are located mainly in the US and China. It is expected that universities in China, the UK, and Australia are more likely to enter the top ten in the future. In response, we need to pay attention to universities arising in the AI field, seek various universities to cooperate with, continuously develop measurement models for AI research indices, and build monitoring systems.
