Quad AI: Assessing AI-related Collaboration between the United States, Australia, India, and Japan

Abstract

Through the Quad forum, the United States, Australia, Japan and India have committed to pursuing an open, accessible and secure technology ecosystem and offering a democratic alternative to China’s techno-authoritarian model. This report assesses artificial intelligence collaboration across the Quad and finds that while Australia, Japan and India each have close AI-related research and investment ties to both the United States and China, they collaborate far less with one another.

Voices of Innovation

“…Previous CSET reports have shown that AI researchers in the United States tend to collaborate more often with researchers in other countries, by comparison with the international collaboration rates of their counterparts in countries such as China and India. 29 Overall, the top 100 U.S. AI researchers in the most-published (45 percent), most-cited (47 percent), and highest-h-index (49 percent) categories did not differ meaningfully from one another in their average rates of international collaboration on AI papers. Moreover, the top U.S. AI researchers' rates of international collaboration are similar to the overall U.S. rates of international collaboration on AI papers (48 percent).…”

“…Moreover, the top U.S. AI researchers' rates of international collaboration are similar to the overall U.S. rates of international collaboration on AI papers (48 percent). 30 From 2010 to 2021, the leading U.S. AI researchers-regardless of background or country of origin-collaborated most frequently with researchers from China, as shown in Figure 6. More specifically, 82 of the 217 top AI researchers in the United States (38 percent) coauthored at least one paper with counterparts in China between 2010 and 2021.…”

Chinese AI Investment and Commercial Activity in Southeast Asia

“…Our analysis builds on prior CSET research in cross-border AI investment using financial data from Crunchbase. 18 No financial database covers the entire investment market perfectly, and Crunchbase data has limitations, including instances where certain information is undisclosed or missing. Where transaction value is undisclosed, we create an estimated value.…”

Financing “The New Oil”: Assessing AI Investment in Israel and the Broader Middle East

“…In the previous example, then, there would be no way to know how much capital was attributed to the American investor; only the total funding raised for that round. 17 Finally, investors are sometimes reluctant to disclose the value of investments, and there is some missing data on how much capital was actually raised by a given AI company. In these situations, we estimated transaction value using a multistage estimation process where we assigned each round the median amount for funding rounds of the same investment stage, target country, and year.…”