The United States holds a leading position in the realm of artificial intelligence, but recent findings indicate that China is significantly accelerating its AI development.
“China is advancing at an incredible pace, and if the US steps back, we could easily close the gap,” remarked Gregory Allen, a senior advisor at the Wadwani AI Center and the Center for Strategy and International Studies.
As per the AI Index Report from Stanford, in 2024, the US had 40 noteworthy AI models, while China had 15 and Europe managed just 3.
David Sacks, the White House’s AI and Crypto Czar, expressed, “Achieving victory in the AI race would mean the US maintains its dominant role on the global stage throughout the 21st century. Since the launch of DeepSeek, China’s AI model development has spanned only three to six months ahead.”
Sacks pointed out that the AI model layer is just a fraction of the competitive landscape, yet China has exhibited remarkable advancements in that area. AI models undergo various assessments, such as language proficiency, mathematical problem-solving, general reasoning, and coding skills. Over the past few years, there have been significant improvements, including an 18.8% increase in general reasoning (MMMU) and a 48.9% leap in advanced science (GPQA). Notably, coding accuracy (SWE Bench) surged from 4.4% in 2023 to 71.7% in 2024.
According to Allen, “AI excels at enhancing its performance across different assessments. It’s akin to standardized testing—do we gauge a student’s overarching intellectual capability or just their preparedness for a test? This duality reflects in AI as well.”
Experts have raised alarms over how swiftly China’s AI models have progressed. At the close of 2023, US systems performed 17.5% in the Language Understanding Test (MMLU), a gap narrowed to just 0.3% by 2024. Similarly, in general reasoning (MMMU), the US advantage decreased from 13.5% to 8.1% during the same timeframe. In the more sophisticated language assessment (MMLU-PRO), China’s DeepSeek-R1 outperformed the American-developed counterpart.
Allen added, “When we talk about the US’s position in global competition, the more telling metric is the worldwide adoption of technologies.”
Interestingly, a recent Fox News poll revealed that many registered voters express a lack of confidence in the government’s capability to effectively regulate AI, with only 38% optimistic about proper regulation.
In 2019, the US began regulating access to high-tech semiconductors, a critical aspect of AI technology.
According to Sacks, “Nvidia represents the US giant in this arena, while China employs Huawei as its top contender. That situation works to our advantage since Nvidia is approximately two years ahead of Huawei. If the US’s tech stack becomes the global standard in five years, we could dominate the AI landscape. Conversely, if Huawei claims 80% of the market, we risk losing.”
Export limitations on companies like Huawei have curtailed access to Nvidia’s cutting-edge chips, pivotal for developing AI technologies. Allen stated, “Without stringent export control measures from previous administrations, AI’s top supercomputers might already leverage American technology.”
Some specialists argue that these regulatory restrictions could inadvertently enhance China’s self-sufficiency, leading them to bypass US technology. Interestingly, the Trump administration recently permitted Nvidia chips back into the Chinese market to counteract Beijing’s swift semiconductor advancements.
In discussions about competition, Sacks emphasized, “Setting the standards means we must scrutinize every company. Every AI developer globally relies on the US technology framework.”
On the semiconductor front, Taiwanese manufacturers currently lead the market, though they aren’t American. Sacks noted that they remain several years ahead of China’s SMIC, further widening the gap.
Concerns linger regarding Taiwan’s geographical proximity to China. Allen reflected on an incident where proximity to a construction site forced significant adjustments due to an impending threat, illustrating the potential risks of a military conflict over Taiwan and its consequences for high-tech advancements.
