This article was originally published on Just Security.
The global competition over artificial intelligence is increasingly framed in stark and dramatic terms, often compared to the Manhattan Project, a new arms race, or a moonshot project requiring incredible resources to attain a difficult, if not, impossible goal. These analogies all suffer from a common flaw: they point us toward the wrong goal. AI is not a discrete project with a clear endpoint, like building a nuclear weapon or landing on the moon. It is a long-term, society-wide effort to develop powerful tools and ensure their benefits reach classrooms, battlefields, factories, and start-ups alike.
The country that leverages advances in AI to establish and maintain substantial economic and military advantages will not necessarily be the one that develops the most advanced models in the shortest amount of time. The United States is not going to win the AI race against China, for example, simply because U.S.-made OpenAI models beat Chinese-made DeepSeek or Kimi K2 models on capability benchmarks. Instead, countries that learn how to bridge the gap between invention and widespread societal adoption will reap the most crucial benefits in the long term.
Read more on Center for a New American Security – CNAS
