Opinion: Why China won’t win the AI war against America anytime soon

Photo credit: Franck Veschi

Joel is a TIA Star Contributor and publishes exclusive, high-value content that serves the Asian tech community. Read more from TIA Star Contributors here.

Up until now, China seems to be doing a good job of catching up with the US-led AI industry. Former Google CEO Eric Schmidt predicts that China will overtake the US to become the dominant global force in AI by 2030. But that’s of little surprise to some, as China’s government has poured massive resources into what they agree will be a critical and decisive technology.

Up to a point, I believe that hard dollar (or renminbi) investment will be the guiding factor in determining the prowess of a country’s AI industry. In addition to tremendous public and private sector investment, China has lots of data, lots of brains, and with its big domestic market, lots of potential customers.

Through such a lens, the argument in favor of China as tomorrow’s AI superpower sounds plausible. But to help us make truly accurate predictions, I suggest that we don’t depend solely on quantitative factors like investment numbers and the amount of data.

Quality over quantity

It’s true that the Chinese use their phones to pay for goods 50 times more often than Americans, and food delivery orders are 10 times greater than in the US. All of this data is therefore available for training AI models in China. Indeed, size and quantity do open the door to great opportunities.

The March edition of The Economist argued, “It is entirely natural for a continent-sized, rapidly growing economy with a culture of scientific inquiry to enjoy a technological renaissance.” The article continues, citing the number of AI scientists in the country and its over 800 million internet users as advantages.

Despite these, however, it remains hard to argue that China will take over the US on AI in terms of quality in the foreseeable future. Here are four key reasons why I believe this.

Lack of global appeal

China, like other Asian countries, does not have enough experience to develop user-friendly software services with global appeal in the way that Google, Facebook, and Amazon do.

In terms of functionality, China has many superior software services, but they are struggling to expand to the global market beyond China. Korea and Japan also face similar challenges.

A lot of datasets don’t guarantee high-quality AI

Data is just one resource. Ultimately, as we see clearly with existing translation services and their shortcomings, inference is more important. By inference, I mean the ability to reason independently, have original ideas, and problem-solve in situations where not all the data is available. This is a key element of machine learning systems that will prevail in the future.

A lot of “brains” don’t guarantee high-quality AI either