Last month, Scale AI CEO Alexandr Wang took a daring step by placing a full-page advertisement in the Washington Post, directed at President Trump, with the message that “America must win the AI war.”
This statement stirred a range of reactions, particularly during Wang’s recent appearance at the opening night of Web Summit Qatar. During the event, when Axios’s Felix Salmon, who was conducting the interview, asked the audience how many supported Wang’s stance, he observed only two hands raised in agreement. Conversely, a significant majority raised their hands in disagreement.
In response, Wang was asked to justify his viewpoint. According to Wang, artificial intelligence is poised to fundamentally transform national security. He shared his background, noting that he was raised in Los Alamos, New Mexico, known as the “birthplace of the atomic bomb,” and that both of his parents were physicists who worked at the Los Alamos National Laboratory.
Wang expressed his perspective on the situation as a competition between the United States and China. He voiced concerns that AI might enable China to surpass the military superiority of “Western powers,” a concern that led to the publication of the full-page advertisement.
Wang’s comments reflected the rhetoric increasingly being used by defense technology startups and venture capitalists, who are advocating for greater autonomy in AI weapons. They are highlighting China as a potential threat, suggesting a scenario where China might deploy fully autonomous AI weapons while the U.S. remains hindered by the necessity of human decision-making before taking action.
Apart from speculating on other nations’ military capabilities, Wang emphasized the importance of choosing between the U.S. and China for foundational large language model (LLM) development. He portrayed this as a two-player race, not mentioning other competitors like France’s Mistral. He argued that U.S. models incorporate principles of free speech, whereas Chinese models align with the viewpoints of a communist society.
There is evidence that many popular Chinese LLM models incorporate government censorship. Additionally, there are ongoing concerns over potential Chinese government backdoors in data collection within these models.
Wang’s concerns about government influence in AI coincided with Scale AI’s announcement of an agreement with the Qatari government. Announced on Sunday, Scale AI will assist Qatar in developing 50 AI-powered government applications, spanning areas such as education and healthcare.
Scale AI is primarily known for employing a large number of contract workers, often based outside the U.S., to assist in training AI models. The company collaborates with entities including Microsoft, OpenAI, and Meta, covering most major U.S. foundational models. It also offers various products, such as an AI data engine and AI applications, some of which are designed for the defense sector.
The explicit pro-American rhetoric is likely advantageous for Scale AI in its dealings with Department of Defense customers. However, the Web Summit discussion also highlighted that many individuals express discomfort with the idea of the U.S. acquiring dominant AI capabilities.
