Words often have different meanings in the business world. A “Unicorn” for example is not a magical horse with a horn projecting from its forehead, but rather a startup that’s valued at more than a billion dollars.
A phrase that’s cropping up a lot in Chinese media these days is “National AI Team (人工智能国家队).” While this may suggest a squad of scientists heading off to some sort of AI Olympics, it rather refers to a new class of AI companies that are either backed by national institutes or closely integrated into government-funded programs.
Why are they calling themselves the National AI Team?
CloudWalk, a spinoff from Chongqing Institutes of Green and Intelligent Technology (CIGIT) of the Chinese Academy of Sciences (CAS), is a typical National AI Team member. Founded in 2015, the company specializes in computer vision and machine learning.
In China, core industries such as public security, banking, political communication, civil aviation, and urban transport are being overhauled through the concerted application of various AI techniques. Security bureaus for example want object tracking systems to help spot criminal activity and hunt fugitives, while banks are using facial recognition algorithms to provide customers with almost-instant access their bank accounts. The government owns or controls most of these industries, and they are rigorous in selecting their technical partners.
“If a bank’s system goes down for two hours, its governor will have a problem; if it doesn’t work out in four hours, he will write a report; if it doesn’t work out in eight hours, it’s going to be a serious incident, the bank’s rating is bound to drop, and the bank may even be closed,” CloudWalk Founder and CEO Xi Zhou (周曦) told Synced.
While a private company’s best way to attract government clients is robust technology or products, being on the National AI Team can also give it an advantage in core industries. Over 100 banks have adopted CloudWalk’s facial recognition technique, making it the largest provider of the tech to China’s banking industry. The company also develops and deploys facial-recognition-enabled surveillance cameras in over 80 percent of domestic airports.
“I was not aware of this advantage until I founded CloudWalk,” says Zhou. “Obviously the state-owned banks and public security departments have more faith in our technologies and products because we come from the Chinese Academy of Sciences.”
What if a startup does not enjoy a relationship with a state-owned institute? They might also look for investment from a state-owned venture. China’s computer vision startup Face++ announced financing of US$460 million last November led by the China State-Owned Capital Venture Investment Fund, while the Guangzhou Municipal Government injected US$300 million into CloudWalk’s Series B funding round.
The Chinese Government owns a large-scale database, which is one of the most important components for developing AI applications and products. Li Xu (徐立) is the CEO of SenseTime, a company that uses deep learning to replicate tasks performed by the human visual system. In an interview with Quartz, Xu boasted that his company has a database of over two billion images. Much of that data comes from various government agencies.
SenseTime raised US$410 million last June, including a B2 round led by Sailing Capital, whose major shareholder is the large state-owned financial holding company Shanghai International Group. SenseTime is now the world’s most valued AI startup. The injection of the state-owned capital is expected to help SenseTime secure more government contracts. At present, 30 percent of SenseTime’s clients are government-related.
National AI Team aims for AI standardization
While National AI Team status may be an “express lane” to government deals, the role also comes with responsibilities. Last year China issued an ambitious policy blueprint calling for the nation to become the world’s primary AI innovation center and aiming to build a domestic industry worth some US$150 billion by 2030. The government is counting on its National AI Team to put that plan into practice.
Over the past five years, the world has seen an explosion of activity in deep learning in both academia and industry all around the world. While AI is energizing industries, the lack of industry-wide standards for research and development is creating a number of problems: How to correctly access users’ data without violating privacy? How to qualify a dataset for use in training AI algorithms? How to ensure the safety and precision of an AI application?
Carlos E. Perez, author of the book Artificial Intuition and Founder of Intuition Machine, stressed the need to standardize the best practices of developing deep learning in his medium blog. “In conventional software development, we have a more mature conceptual framework that has evolved over time. Deep Learning introduces new kinds of requirements, so we need to understand what these are and what standardize the class of tools needed.”
The Chinese Government is pushing hard for AI standardization. Last year it released its Standardization of AI Helps Industry Development along with a blueprint for further AI integration with the economy. This year it followed up with the Artificial Intelligence Standardization White Paper.
In March 2017 the National Development and Reform Commission launched the Public Service Platform for Basic Resources (人工智能基础资源公共服务平台) with a heavy focus on AI. This platform will be built over the next three years and aims to deliver:
- a large-scale computing cluster;
- an open-source database with a capacity of not less than five petabytes, and at least five million pieces of data resources;
- a smart system that can automatically label data and conduct smart analysis;
- image recognition with 93 percent accuracy, Mandarin character recognition with 90 percent accuracy, and speech recognition with 95 percent accuracy.
Prominent National AI Team players Baidu, Tencent, iFlytek, and CloudWalk were designated to take the lead on the Platform’s development.
China’s voice technology giant iFlytek was founded by alumni of the University of Science and Technology of China, a national research university, while Baidu and Tencent have dominated the country’s tech scene for years.
CloudWalk spokesperson Xiaolong Fu told Synced, “AI is closely related with data and privacy, so National AI Team members can help ensure information security. Meanwhile, we also want to ensure that our intellectual property remains in our own hands.”
China’s quest to develop cutting-edge AI technologies provides a world of opportunities for companies on the National AI Team, particularly with the increasing number of government initiatives and state-run projects in the field. The Team may also spur development and deployment of improved AI resources for academic, industry and public use.
Journalist: Tony Peng | Editor: Michael Sarazen