Jeffrey Ding in Iowa City: Diffusing AI across the economy, not just chasing breakthroughs

Jeffrey Ding, an assistant professor of political science at George Washington University, told an Iowa City audience that the strategic competition over artificial intelligence will hinge less on who first creates frontier models and more on which countries can embed AI broadly across industries.

Ding laid out a contrast between two accounts of technological competition. "Leading sector" theory focuses on a country's ability to monopolize a new, fast-growing industry. By contrast, his research centers on general purpose technologies — innovations that spread across many sectors. "I think this image illustrates that we have a pretty confused ... understanding of what US China competition and AI would even look like," he said, arguing that AI resembles historical GPTs such as electricity and the steam engine.

Ding said GPTs require different institutional responses because they depend on a broad base of applied engineering skills and coordination across application sectors. "It's about average engineers, not about the best and the brightest experts," he said, stressing the role of universities and regional institutions in training a larger pool of practitioners able to apply AI in manufacturing, education and other fields.

To support the claim that diffusion — not monopolized innovation — shaped past power transitions, Ding pointed to the second industrial revolution. He described machine tools and standardized, interchangeable parts as a central GPT that spread through U.S. industry, enabling mass production and productivity gains. "When your bicycle breaks down, instead of remaking the entire bike, now the parts are standardized so the company can just replace the part that broke down," he said.

Applying that history to today, Ding presented evidence that the United States has a substantially wider bench of institutions capable of training AI practitioners: "The U.S. had 159 of those universities, China only had 29," he said, referring to his count of institutions with at least one researcher who published in established AI conferences. He argued that beyond elite research universities, it is the regional and state university systems and the strength of industry–academia linkages that will determine how quickly AI diffuses across the economy.

During audience questions, Ding addressed whether open-source models such as the recently released DeepSeek improve diffusion. He replied that an open-source ecosystem can facilitate adoption but that sectors differ in how much fine-tuning they need and in the costs of using open vs. closed systems. On public reluctance to adopt new technologies, he recommended safety demonstrations, prudent regulation and civil-society transparency measures to build trust.

The lecture also touched on environmental constraints: Ding noted growing demand for data centers and chip clusters and said some tech firms are negotiating energy solutions, including small modular nuclear reactors, to meet power needs for training large models. He cautioned that if scaling laws require ever-larger compute, energy availability will shape the trajectory of model development.

Asked whether AI might be a bubble, Ding said he was not a technological foreteller but cited patent-citation diversity as an indicator that AI has a credible case to be a general purpose technology. He also addressed whether China's political system makes diffusion incompatible with its goals, saying Beijing wants productivity growth and broad diffusion but that centralization can impede organic, economy-wide diffusion processes.

Ding closed by noting his book, Technology and the Rise of Great Powers (Princeton University Press), is available and invited further questions. The Iowa City Foreign Relations Council and the University of Iowa co-hosted the event at the Iowa City Public Library.