After all, how dependent is China on US AI technology?

The United States of America (USA) is erecting high barriers to China’s growth, trying to prevent its success. One of the most critical areas is Artificial Intelligence (AI), whose American technology has been kept away from the Asian country. But, after all, how dependent is China on the USA?

Yesterday, we reported that the USA plans to increase protectionism regarding AI models developed in the country and which power popular chatbots, such as ChatGPT. The objective is to safeguard its technology from powers such as China and Russia.

However, China has advised companies to avoid foreign technology, motivating them to develop their own alternatives, in order to strengthen the country – which has been harmed by measures imposed by several governments.

Given this scenario and the constant attacks (mainly American) against the Asian country, Reuters tried to understand, here, how dependent China is on the USA in terms of AI and the impact of the Joe Biden administration's plans.

First, do China's AI models use American technology?

Apparently, yes. However, the idea seems to involve developing entirely Chinese models, without depending on technology provided by American companies.

The US wants to prevent the export to China of patented or closed-source AI models, whose software and training data are kept secret. However, open source models are not, from the outset, covered by the potential American barrier.

In any case, the Chinese Academy of Artificial Intelligence, a high-level research laboratory, admitted in March that the majority of Chinese AI models developed internally were, in fact, built based on Meta's Llama model and that this represented a fundamental challenge for the development of AI in China.

At the time, the laboratory told Chinese Prime Minister Li Qiang that China “seriously lacks autonomy” in this area.

In fact, in November 2023, 01.AI, one of the best-known AI unicorns in China, founded by a former Google executive, Lee Kai-fu, faced controversy after AI engineers discovered that its IA, the Yi-34B, had been developed based on the Llama da Meta.

Despite this dependence, numerous technology companies such as Baidu, Huawei and iFlytek have been working to develop their own completely proprietary AI models. Some of them claim that their proposals could be as capable as OpenAI's latest GPT-4 model in several areas.

Next, what is China's stance on US AI models?

In line with the warning left by the Chinese Academy of Artificial Intelligence, Chinese entities have highlighted the need for the country to develop its own “controllable” AI technology. These reservations arise in line with an order from Chinese President Xi Jinping to develop technological self-sufficiency.

Additionally, the country has been proactive in implementing regulations on the use of generative AI, requiring services to have government approvals before reaching the public. Since January, China has approved more than 40 AI models for public use, and none of them were foreign AI models.

According to Reuters, the Chinese government has not encouraged the use of American AI technology. Instead, it has sought to motivate Chinese companies to work on their own technology, reminding them of the gap that (still) separates their proposals from those of the United States.

This need for emancipation was reinforced in February, with the state-owned newspaper China Daily recalling that ChatGPT “could provide assistance to the US Government in its dissemination of disinformation and manipulation of global narratives for its own geopolitical interests”.

Briefly

Although China still depends on US AI, whether in terms of hardware or software, and sees the measures applied and planned as an obstacle to its own development, the truth is that the country seems to be striving for independence, seeking to develop solutions themselves as good, or even better.