The competition between the US and China over artificial intelligence (AI) is reshaping global power dynamics, with significant implications for both nations and the world at large. Key developments in this race include the ongoing Ukraine war, which has underscored AI's military potential, and the launch of OpenAI's ChatGPT in November 2022, which has positioned the US as a leader in generative AI [6d05ac59]. However, while the US excels in generative AI, China has been outpacing the US in AI patents since 2021, indicating a robust innovation pipeline [6d05ac59].
Recent reports suggest that US efforts to restrict China's AI advancement are deemed unlikely to succeed. A Washington-based think tank, the Information Technology and Innovation Foundation (ITIF), highlights that China is effectively circumventing US export curbs and fostering local innovations [00cac329]. The US has imposed strict export controls on advanced AI chips to China, but these measures have not significantly hindered China's progress [00cac329].
China has made substantial investments in generative AI since late 2022, with state-backed funds investing $184 billion in 9,623 local AI firms from 2000 to 2023 [00cac329]. Additionally, China published over 234,000 AI research papers from 2017 to 2022, surpassing the US's 172,600, and filed six times as many generative AI patents as the US from 2014 to 2023 [00cac329]. This trend indicates a growing dominance of Chinese firms in the generative AI patent landscape.
Experts now estimate that China is approximately one year behind the US in developing large language models (LLMs). Roey Tzezana from Tel Aviv University states that the gap in text generation LLMs is about six months to a year, with no Chinese LLM having surpassed OpenAI's GPT-4 [16114a59]. OpenAI's latest model, o1, released on September 12, 2024, excels in reasoning tasks and is described by Yang Zhilin from Moonshot AI as a paradigm shift due to its reinforcement learning capabilities [16114a59]. Despite this lag in text models, Chinese firms are reportedly advancing in edge AI and text-to-video generation, indicating a shift in focus [16114a59].
The US government has committed over $12 billion to AI research and development over the next five years, reflecting its strategic priority in this sector [6d05ac59]. In contrast, China faces significant challenges in chip production and data sourcing due to stringent censorship and regulatory hurdles, which could hinder its AI advancements [6d05ac59].
Recent statistics reveal that 57% of elite AI researchers are based in the US, while 26% are from China, highlighting the concentration of talent in the US [6d05ac59]. This talent pool is crucial as Goldman Sachs predicts that generative AI could boost global GDP by 7% over the next decade, signaling the economic stakes involved in this technological race [6d05ac59].
However, concerns have been raised regarding potential unemployment in poorer countries as AI technologies advance, suggesting that the benefits of AI may not be evenly distributed [6d05ac59]. Importantly, the race may not be zero-sum; there are opportunities for global wealth increases that could benefit multiple nations, provided that collaborative frameworks are established [6d05ac59].
In the context of this competition, Chinese AI start-ups are increasingly viewing overseas expansion as essential for their survival and growth. A recent report highlighted that 103 out of 1,500 active AI companies globally are from China, emphasizing the country's significant presence in the AI landscape [5cbdc653]. Major players like Alibaba and ByteDance are launching initiatives aimed at international markets, reflecting a broader trend among Chinese firms adapting to local regulations and market demands [5cbdc653].
Following OpenAI's recent preview of its new language model, OpenAI o1, on September 19, 2024, Chinese AI start-ups are actively seeking fresh opportunities. Founders of prominent Chinese firms, such as Yang Zhilin of Moonshot AI and Jiang Daxin of StepFun, have emphasized the importance of scaling through reinforcement learning and how OpenAI o1 will drive innovation in foundational models [96b250bd].
Despite facing US trade sanctions that limit access to advanced semiconductors, only about 10 AI start-ups, including 'AI tigers' like Moonshot AI, Baichuan AI, Zhipu AI, and MiniMax, are expected to invest heavily in reinforcement learning [96b250bd]. As the global AI race intensifies, the need for Chinese companies to establish a foothold in international markets becomes increasingly critical. This shift not only reflects the competitive pressures within China but also the strategic importance of global expansion for sustaining growth and promoting Chinese culture [271c5173][5cbdc653].
Amid these developments, the US-China rivalry is complicating global efforts to regulate AI. China opted out of a non-binding international declaration on AI and nuclear weapons control at the REAIM conference in Seoul on September 10, 2024, which was endorsed by over 60 countries, including the US and Ukraine [aadb6567]. Observers note that this reflects China's dilemma in balancing military AI rivalry with the US and its desire for a greater role in global governance. Experts suggest that China is cautious about commitments that could limit its future options [aadb6567]. The US-China strategic rivalry further complicates global AI regulation efforts, as evidenced by China's abstention from a UN resolution on AI in autonomous weapons in November 2023 [aadb6567]. Although the US and China held their first AI dialogue in Geneva in May 2024, experts call for limited agreements on specific AI applications amid ongoing geopolitical tensions [aadb6567].