China risks a catastrophic loss in competitive advantage to the US in artificial intelligence (AI) if it does not invest more in basic research, according to Peking University professor Wang Liwei.

The next technology breakthrough is more likely to happen in America than in China because the latter trails in basic academic research, said Wang, who specializes in machine perception and is a member of the China Computer Federation’s AI and pattern recognition committee. With a tech war raging, China could find itself cut off from the latest advancements in the technology, he said.

RNG74Y Cybernetic brain, deep machine learning and artificial intelligence concept 3D illustration. (Picture: Alamy)

“I think the ability to make such a breakthrough represents the real level of AI development of a country,” Wang said in an interview in Shanghai last week. “You can imagine what will happen if it’s in the US, not in China.”

China is engaged in a race with the US in AI development, with China’s State Council issuing a three-step plan in 2017 to make the country a global leader in the technology by 2030. It is a battle fought in the laboratories in universities and companies as much as in real-life application and strategic thinkers in both countries see mastery of AI as a key to unlocking future economic and military gains.

China lags behind the US in the global talent needed to power AI research and development. China has fewer than 30 universities with AI labs. Forty percent of data scientists in China have less than five years of working experience, while in the US, more than half of the researchers have more than 10 years of experience. China also has about half of the global AI talent compared to the US, according to the 2018 Deloitte AI report.

Even so, China has made gains in AI in recent years largely on the back of improving on open-source machine learning libraries such as Google’s Tensorflow and Facebook’s Pytorch, which developers grant users the right to study, change and distribute the code for collaboration, according to Wang.

Today's AI is still in an infant stage compared to the ideal artificial neural network, said Wang. Progress in deep learning, a form of AI that mimics the workings of the human brain in processing data for use in decision making, has slowed in the past two years and may be reaching a ceiling, Wang said.

That is why breakthroughs are likely to be grounded in fundamental research, he said.