The Geopolitics of AI Middle Powers

There are many open questions around the importance and strategic role of middle powers* in the U.S.-China AI race. I believe that meaningful progress could be made on these questions by conducting and publishing a series of quick analyses and posting them on twitter or blogs – something along the lines of Lennart Heim’s twitter threads. I believe that the answers to these questions could inform the strategic decisions of governments that are developing AI strategies.

Some potential questions for analysis include:

Question 1:

How do Russian experts expect that Putin would react to the U.S. and/or China developing advanced AI systems? Potential method: Informal interviews with 3-5 Russian national security experts. Mentee’s potential role: Do background research on the Russian government’s perspective on relevant factors such as 1) Russian reactions to being “behind” in key technology developments

Question 2: What are “budget” interventions that middle powers can take to have an outsized effect on AI safety without having to spend much money or political capital? Potential method: Literature review on the historically most impactful roles that foreign countries have played on U.S., Chinese, or other programs in terms of safety. (e.g. whistleblower protections programs)

*Note: When I say “middle powers” I’m referring to states that are not the U.S. and China that still have relevance to the development of advanced AI systems – as funders, semiconductor supply chain countries, potential regulators, etc.

My personal shortlist of middle powers includes: Brazil, Canada, EU, France, Germany, India, Indonesia, Israel, Japan, Kenya, Mexico, Nigeria, Netherlands, Saudi Arabia, Singapore, South Africa, South Korea, Taiwan, United Arab Emirates, United Kingdom, Vietnam.

Selection criteria (applied to the entire country list): Each country does at least one of the following hosts an influential AI company or company in the semiconductor supply chain, or commands large AI/compute investments, or operates a notable AI‑governance system, or is building new, notable AI or compute capacity or is a key research node for AI and/or semiconductors or a key source of the critical minerals needed to source the AI chips or a key supplier/source of energy and/or capital that is attractive/essential to the frontier labs

There is little research on the strategic role of middle powers in the geopolitics of advanced AI. This includes potentially key topics, such as 1) what checks and balances there could be on leading AI companies from foreign investors and governments or 2) which countries could increase competition or risky AI deployment dynamics in the U.S. or China. I believe that doing short analyses relevant to these questions could improve the state of discussion around the geopolitics of advanced AI.

Mentees would ideally play the role of a researcher and writer. Some mentee work may be to prep the mentor's own work -- such as preparing interview questions that the mentor can experts in her network in order to answer the short analysis questions.

While this project is focused on quick and short analyses, we can always adjust to spend more time on impactful questions that we're making progress on!

-Clear writing and communication in the style of reasoning transparency. -Interest in understanding the details of a wide variety of topics in AI geopolitics. -Clear strategic thinking around transformative AI strategy. Specifically, the ability to identify key questions to pursue and determine what information could be used to answer those questions. -Experience conducting literature reviews and interviews (nice but not necessary).

NA