Former Google CEO Says US Is 'Two or Three Years Ahead' Of China in AI For These Reasons

Former Google CEO Eric Schmidt suggested that the U.S. and China are at different stages of A.I. development, with the U.S. becoming more advanced. Schmidt outlined four reasons China may need help to match the U.S. in A.I.

In a Bloomberg interview on Tuesday, Schmidt declared that the U.S. holds a solid two-to-three-year advantage over China in A.I. That's a significant lead in this field, he noted. Following his tenure as Google's CEO (2001-2011) and chairman (until 2015), Schmidt has deviated his focus to A.I. investment.

Among his ventures is Anthropic, whose A.I. chatbot Claude recently debuted on iPhones. Schmidt further solidified his involvement in A.I. by chairing the Department of Defense's Innovation Board (2016) and the National Security Commission on Artificial Intelligence for three years.

Schmidt warned that the U.S. could only secure victory in the A.I. race by avoiding complacency and capitalising on its current lead. Meanwhile, China's A.I. development is accelerating.

In the first half of 2024 alone, authorities approved over 40 A.I. models, including a surge of 14 large language models (LLMs) released for public use within a single week. Baidu, China's search engine giant and Google equivalent, is at the forefront of this rapid progress.

Schmidt pinpointed four areas where China falls behind in the race for A.I. dominance.

Chip shortages

Schmidt attributed China's chip struggles to global shortages and restrictions on high-speed chip imports, particularly Nvidia chips, imposed by the Trump and Biden administrations. "They're certainly angry about that," he added.

It is no secret that these chips are the brains of modern technology, and their power is crucial for advancing Artificial Intelligence. Political and economic competition between the U.S. and China has motivated the U.S. government to invest in domestic semiconductor production, reducing reliance on foreign sources.

The U.S. Department of Commerce's Advanced Computing Chips Rule (November 2023) restricts China's ability to import high-tech A.I. chips from American companies, limiting China's access to cutting-edge A.I. technology.

In March this year, the U.S. contemplated sanctions on Chinese semiconductor firms tied to Huawei. This followed Huawei CEO Ren Zhengfei's claim last year that the company had successfully replaced over 13,000 US-made components in its products.

Lack of Chinese Material to Train A.I. Models

Schmidt pointed out a key challenge for Chinese language models on CNBC: The limited availability of training data in Chinese. He noted that English dominates the internet, research papers, and books - the primary data sources for training LLMs.

According to Schmidt, this gives English models a significant advantage as they have a much larger pool of information to learn from. "That's why English is so strong in these large language models," Schmidt said.

Schmidt further noted on CNBC that the dominance of English in training data for large language models presents a twofold challenge. Firstly, limited data is available in other languages, hindering their development.

Secondly, this English bias can lead to misunderstandings and misinterpretations when these models are applied to other languages.

Lack of Funding

Schmidt highlighted contrasting situations for China and the U.S. in foreign investment and venture capital. According to Schmidt, while China faces a significant reduction in these areas, the U.S. has seen a boom.

He also pointed out that China's economy has declined recently, and deflationary pressures persist. That analysis aligns with what experts observed in late 2023. China experienced its first foreign investment deficit in November of last year.

This coincided with ongoing tensions with the U.S. and a shift away from Chinese business involvement by some Western countries.

Not Focusing On The Right Areas

Schmidt also criticized China's tech industry for prioritizing for-profit applications. Although he acknowledged the potential success of these individual applications, he suggested a broader issue exists.

The main criticism is the need for a greater focus on building overarching platforms, which Schmidt argues fosters long-term growth in the tech industry.

"Three or four of the top apps in America are, in fact, of Chinese origin," Schmidt said. "But at the moment, the leadership is U.S."

Despite the success of consumer applications like TikTok, China needs help with foundational A.I. models, according to a CNBC report. On CNBC, Eric Schmidt expressed strong national pride, stating, "We should be very proud to be here."

He noted that the U.S. has pioneered this future, driven by A.I., quantum computing, and other emerging technologies. "If we do it right, we have a shot of actually dominating the world for the next 10 or 20 years," he said.