By Nina Xiang*
Over the past several years, we have witnessed the blossoming of China’s artificial intelligence industry. Thanks to generous government policy support, and enthusiastic entrepreneurs and venture investors, China commandeered a position that it has not occupied for centuries: that of a global leader in an emerging critical technological field.
But as we entered 2019, whispers of an AI “winter” began to emerge. The venture capital market has indeed dialed back. The number of venture financing deals in China’s AI sector as of mid-June 2019 stood at 131, with a total deal value of $5.6 billion, as compared with 496 venture deals and $15.7 billion in total deal value in 2018. The number of mega-rounds – those venture deals greater than $100 million – fell from 26 such deals in 2018 to 4 so far in 2019.
At the same time, the bottlenecks of deep learning – the driver of the current wave of AI enthusiasm – are more frequently discussed. The lack of explainability, lack of ability to reason, the need for a large amount of data for training, and the challenges of solving real-world problems are constraining the technology’s commercial applications.
The spillover of the US-China trade and technology frictions is adding additional uncertainties to the industry’s future. The Chinese AI industry advanced during the heyday of free academic exchanges, thriving global open-source communities, and free talent flows between the United States and China. The continuation of these elements can no longer be taken for granted going forward.
As such, artificial intelligence in China has entered a stage of rationalization and recalibration, leading to greater unpredictability. China’s AI industry is poised at an inflection point. The question of how China’s AI industry will be shaped in the future becomes more significant, yet less visible.
Ten industry experts have offered their outlook for the industry for the coming decade. Despite the aforementioned challenges, these experts are overwhelmingly optimistic about AI’s continued advancement and commercialization in China. Underpinned by Beijing’s unwavering commitment to advance Chinese technology, AI is likely to continue expanding its disruptive powers across industries more deeply and rapidly in China than in other countries. Here, we highlight some of their predictions:
“AI will enhance the capabilities of grassroots-level hospitals and doctors, so that their diagnostic abilities will be on a par with premium-level hospitals and doctors. Our experiments show that for medical image reading on pulmonary nodules, the performance of a junior doctor assisted by AI has no statistical difference from that of a veteran doctor. For events such as strokes, as well as for diseases, the superiority of artificial intelligence in auxiliary diagnosis is particularly prominent. Hence AI can help level the healthcare resource imbalances between urban and rural areas, and between more developed and less developed regions.” – Kuan Chen, founder of Infervision
“We will see the largest explosion of intelligence in human history occurring in the upcoming decade. I expect that within three to five years, the proportion of Chinese companies’ annual revenue driven by AI will exceed 50%. Within five years, the leading companies will have almost completed the shift toward AI driving most of its businesses.” – Wenyuan Dai, founder of 4Paradigm
“Chinese AI companies are expanding their technological tentacles and trying to achieve full-stack, cross-industry capabilities. This is an enormous challenge, testing every team’s resources, capital, talent pool, and business development abilities. In addition, there is a growing consensus that focusing on use case-specific solutions and deep optimization is the passage to the future. AI startups realize that the emphasis on building platforms, as companies did during the internet era, doesn’t work well in the AI era.” – Wei Huang, founder of Unisound
“In the future, intelligent risk-control systems within the financial services sector will follow three major trends. First, in terms of timeliness, it will transition from after-the-fact, stop-loss risk control to a real-time feedback model. Second, in terms of decision-making, it will gradually shift from human manual review and a posteriori strategy, to machines replacing most of the manual work and an automated decision-making process. Third, the balance between risk control and user experience will gradually reach an equilibrium.
In the future, anti-fraud and financial risk-control systems will become so efficient and frictionless that end users will not feel the effort behind significantly enhanced security.” – Tao Jiang, founder of Tongdun Technology
“In the era of AI implementation, personal privacy and data security have always been important issues. We should not view this simply as a regulatory problem. Instead, we should try to fight and solve the challenges that accompany new technology by creating better technology. For example, technologies such as homomorphic encryption and federated learning can protect personal data, as well as enabling data to empower our AI algorithms and improve their efficacy.” – Kai-fu Lee, CEO of Sinovation Ventures
“The boundaries of AI technology are far from certain. Today, AI algorithms are not converging. On the contrary, the gap between algorithms is widening. Many application breakthroughs will take place in the industrial sector. At the same time, the development of AI technology is cross-industry. It can directly enter an industry, breaking the existing structure of the industry and reshape it, allowing the industry to leap forward.” – Chenxi Lin, co-founder of Yitu Technology
“Over the next five to 10 years, the countries and regions where self-driving cars are deployed earlier will see increasing maturity of the technology, while other areas will lag ever farther behind. As time goes on, this disparity will increase. China enjoys certain advantages including diverse and challenging road conditions, strong policy support and high public acceptance. I think China has the potential to be a frontrunner for the advancement of self-driving technologies and the commercialization of the technologies.” – Tiancheng Lou, founder of Pony.ai
“The worst-case scenario is that the world will face a balkanization of its digital and AI-powered future, in which the United States and China – and perhaps other countries as well – each have their own independent IT infrastructure and ecosystem. The division and disjuncture of the physical world will extend into the digital and virtual spheres, greatly inconveniencing users and creating greater uncertainty for our collective future. A more favorable outcome would include the continuation of the status quo with minor disruptions, moving toward accelerated AI applications across industries globally and enhanced international collaboration.” – Nina Xiang, founder of China Money Network
“The application of artificial intelligence will be highly customized to real-world use cases. Whether we are talking about autonomous driving, smart cities or smart retail, the use case determines the type of algorithms needed. China’s future development of the internet of things powered by smart cities, smart transportation, industrial 4.0 and 5G means AI chips need to focus on edge computing, as well as the simultaneous development of chips and algorithms.” – Kai Yu, founder of Horizon Robotics
“Once Chinese AI companies achieve the ability to provide the best AI products globally, they will have wonderful opportunities to expand overseas. But this global expansion will take a long time – at least five years.
Also, because Chinese AI companies may have reached world-leading levels in terms of technology sophistication, they will be able to expand around the world. Unlike earlier Chinese internet companies that focused on expansions in developing countries, Chinese AI companies will not be confined to seeking market share only in emerging markets.” – Xi Zhou, founder of CloudWalk.
*Co-founder, China Money Network
** First published in weforum.org