AI in China: How AI Can Optimize Your Operations

August 7, 2017 Posted by China Briefing Reading Time: 7 minutes

By Gidon Gautel

This is Part 2 of a two-part article on Artificial Intelligence (AI) in China. In Part 1, we discuss foreign investment opportunities within the industry.

Many business leaders across the world, including China, do not yet see AI as a priority; at least 40 percent of enterprises within traditional industries in China do not see AI as strategically important. Similarly, in the West, one study suggests only 43 percent of business leaders are likely to implement any type of artificial intelligence in the next three to five years.

A survey by Forrester has found that, of those businesses within the study who choose not to implement AI, 43 percent said there was no defined business case, 39 percent said they were unsure what AI could be used for, and 33 percent did not have the required skills for implementation.

This comes despite the fact that, according to PWC, 55 percent of near term gains from AI are set to come from productivity increases in labor. These gains will take the form of increased automation of routine tasks, greater devotion of time to higher value work, and through general augmentation of employee capability. All of these are highly relevant to SMEs.

Firms should study the structure and operation of their business, distinguish areas ripe for implementation, and survey the market for available AI services and associated talent to invigorate these areas. Even though AI is in its early stages, and not every business needs to implement immediately, the winners of tomorrow will be those who at least have a plan.

Application program interfaces

The simplest way businesses can start implementing AI into their business is with Artificial Intelligence Application Program Interfaces (APIs), which are readily available on the market.

To give an example, one company, iFlytek, has garnered significant appraisals for intelligent language tools extremely relevant for application in business, despite some delay in the application of Natural Language Processing (NLP) in applied solutions, such as smart speakers.

Many Chinese companies, and even the government, use the company’s services for processes such as transcribing courtroom interactions. According to the Chinese Academy of Sciences, its software beats Apple’s accuracy rate for speech recognition of the Chinese language.

iFlytek currently offers software for industry enterprises that can sift through volumes of recording data, convert this into text, and analyse the content of this new data. Baidu also offers a wide range of APIs, including NLP tools.

To give one example of the strength of such software in customer service, such tools could be used to transform hours of audio of customer service interactions into a text-based format. Going over this data with text and sentiment analysis tools can offer insights into areas such as consumer preferences and current service standards.

Machine learning services

Currently, machine learning (ML) services are another area of AI where there is already an established market with several platforms to choose from.

The predictive prowess of ML techniques allows businesses to benefit from its applications in a wide variety of ways. Machine learning can improve customer loyalty and retention through the analysis of typical user behavior and optimization of the provision of offers and services based on this.

In logistics, more accurate ML models can replace old linear regression models to forecast demand. ML and other AI solutions can aid in fraud prevention by seeking out “unusual” user behavior.

ML also has a breadth of applications in manufacturing. Using data generated by monitoring sensors, manufacturing plants can integrate ML solutions to enhance efficiency. A graphic from General Electric estimates proactive equipment tuning can increase production capacity by up to 20 percent. Predicting production capacity could allow companies to reduce finished good buffers by up to 30 percent.

Machine learning can also reduce quality inspection resources by up to 60 percent. Neural networks can generalize historical quality control data and identify defects while ignoring other features, such as complex geometry or image noise, which are often mistaken for defects by human inspectors.

Baidu is currently rolling out more AI services and useful APIs. However, in the area of cloud-based machine learning as a service, Microsoft, Amazon, and other Western services are often a better option for businesses wanting to make use of advanced infrastructure over the cloud. Microsoft has built up dedicated data center services for Azure on the mainland.

Some firms may want to develop and run models locally on an open source AI platform instead of using cloud-based machine learning tools. This offers a higher degree of customization and allows businesses to easier comply with policies related to data confidentiality and sovereignty, which is important in China, considering its new Cybersecurity Laws.

In terms of open source AI tools, Google’s Tensorflow is arguably one of the best options on the Western market. However, the platform is handicapped in China because users cannot access Google Cloud, through which it is optimized, due to the Great Firewall. Baidu’s own deep learning platform, PaddlePaddle, obviously does not have this problem.

PaddlePaddle boasts several other advantages. Baidu specifically advertises the fact that PaddlePaddle makes it easy to scale heterogeneous computing resources—systems that use more than one type of processor. PaddlePaddle is designed to work well with clusters: several computers working together, and making use of GPU (Graphics Processing Unit) clusters to speed up deep learning can result in faster model training and inference times. Furthermore, the toolkit is simpler to use than its rivals. A translation program on PaddlePaddle requires up to 75 percent fewer lines of code than other platforms.

As Baidu, Alibaba, and Tencent (BAT), as well as smaller players, continue the push to develop AI, more Chinese tools will present themselves for use by foreign businesses. It remains to be seen which services will rise to dominate.

Finding the talent

Developing tools internally will require in-house talent. While there is currently an AI talent shortage in China, Christopher Piazza, Product Marketing Manager for PCs and Smart Devices at Lenovo, is doubtful whether this will serve as any real advantage over local firms for foreign businesses able to source in-house engineers from abroad. According to Piazza, “Collecting data, aggregating it, and pushing that out to a better user experience is relatively boiler plate once the mechanism is in place”.

Thomas Zhang, Group IT Director at Dezan Shira & Associates, adds: “The shortage is for those people who can play lead roles. For general staff needed in the AI industry, I don’t think there will be a problem for local companies. China has so many people and AI is such a hot topic, that this will attract many people shortly.”

Looking ahead, Piazza maintains that firms implementing AI and looking for the talent to do so should be looking for individuals able to leverage AI in order to enhance added value. Piazza said, “Offering end user services and a thought-out ecosystem creates a far more powerful added value, meaning more customer service facing positions in the AI job market that require technical AI savvy, with a sales/marketing approach to real benefit and execution”.

Planning ahead

For foreign businesses wanting to implement AI, China’s rise within the industry should lead to greater competition for the best talent and tools, and increase service options in the market.

AI is a far-reaching field and this article has not further explored the use of AI in autonomous driving, logistics or the financial sector, to name a few. Business leaders in all industries should spend the coming years studying their field for best practices and opportunity, as China continues its AI ascendance.

This is Part 2 of a two-part article on Artificial Intelligence (AI) in China. In Part 1, we discuss foreign investment opportunities within the industry.

About Us

China Briefing is published by Asia Briefing, a subsidiary of Dezan Shira & Associates. We produce material for foreign investors throughout Asia, including ASEAN, India, Indonesia, Russia, the Silk Road, and Vietnam. For editorial matters please contact us here, and for a complimentary subscription to our products, please click here.

Dezan Shira & Associates is a full service practice in China, providing business intelligence, due diligence, legal, tax, IT, HR, payroll, and advisory services throughout the China and Asian region. For assistance with China business issues or investments into China, please contact us at china@dezshira.com or visit us at www.dezshira.com