City of Tomorrow

By staff reporter VERENA MENZEL

IT is often said that the digital revolution will effect radical changes in the way people live and work， hinting at future trends. However in Shenzhen， that future is already happening.

Shenzhen， a city of around 13 million people， last year celebrated its 40th birthday as Chinas first special economic zone. The young metropolis has emerged as a world leader in the ongoing technological revolution， driven by its nascent energy.

For a glimpse into the future， a visit to the citys Ping An Finance Center is a must. At 599 meters high， the headquarters of Ping An Insurance is the second tallest building in China and towers over all the other skyscrapers in Shenzhen. Behind the glittering glass facade， more than 20，000 people are employed， all playing their part in helping to shape tomorrow.

Voice Recognition

Wayne Hu is deputy director of Ping An Smart City Development Office as well as co-president and chief technology officer （CTO） of Ping An Smart City， a subsidiary of the group. Hu said that since its founding in 1988， Ping An has always offered much more than just insurance. The 32-year-old Fortune Global 500 Chinese enterprise recognized about a decade ago that the future will merge finance and technology.

Today， Ping An employs over 30，000 scientists and engineers worldwide， including more than 3，000 leading experts entrusted with research and development （R&D） of emerging technologies. The companys eight research institutes and 58 laboratories are primarily engaged in fields of artificial intelligence（AI）， financial technology （fintech）， health technol- ogy， smart cities， blockchain， cloud computing， the digital economy， and biology.

Traditionally known for providing insurance services by much of the public， Ping An has rocketed into becoming a world-leading hi-tech company in just 10 years. Its numerous innovations will likely play a decisive role in shaping the future world， if the large volume of international patents filed by the company in recent years is any indicator. In the sector of digital health alone， Ping An now ranks second globally in terms of its quantity of published patent applications.

The companys showroom gives an early look at what work and life in tomorrows metropolises may look like through displays of Ping Ans flagship products. A common feature is organic integration of familiar things， with the new possibilities created by new technologies.

An example of this organic integration is how an AI-based customer service system can interact with customers. In a simulated demonstration， an irate customer is firmly but professionally dealt with by an AI service robot with a voice and reactions amazingly similar to a real human.

The robot is able to understand and speak a wide variety of local dialects， and can also automatically adjust its communication strategy in seconds based on customers reactions. According to the public relations staff， a 10-second voice sample in any language is enough for the robot to accurately imitate it and store it in the system. This new technology has revolutionized Ping Ans customer service.

“About 90 percent of over 2 million customer service calls every day are processed through our AI system，” says Hu. “Customers can request to speak to a human at any time， but surveys showed that our customers are more satisfied with the AI system than with human employees. Customers care most about whether or not their problem is solved. It doesnt matter whether a robot or a live person is on the other end of the line，” he added.

Using AI technology in customer service has several advantages. First， unlike humans， robots are not emotion-driven. “With human employees， their state of mind can noticeably affect service quality，” says Hu. Second， the knowledge of human staff is naturally limited. “Our service catalog is extremely extensive，”he continued. “None of our employees could possibly memorize every detail， but the AI system can make a big difference. If the information the customer seeks is in our database， the robot can always provide a precise answer.”

The system has also made several months-long employee training courses obsolete， which has led to significant cost reductions. “We previously employed 120，000 customer service agents，” says Hu. “Today， the staff size has been cut in half. Over the next two years， the number could fall further to roughly 20，000.”

This customer service system has only been used in China to date， but it could be adapted to a wide variety of foreign languages and serve other countries. Such an AI-based voice technology offers a taste of what tomorrows customer service will look like in major cities around the world.

Micro-Expression Analysis

Laboratories in Shenzhen are exploring much more than the human voice. Ping An researchers have also scrutinized body language and microexpressions. Regarded as intuitive knowledge， facial expressions can communicate thoughts and convey consciousness， and AI scientists worldwide are now researching the subject. In Shenzhen， scientists are inching closer to valuable insight into the workings of the brain.

One impressive technology is a micro-expression recognition system which Ping An is using to evaluate credit worthiness. The system can detect micro-expressions by analyzing muscle movements of different parts of the face such as eyes， eyebrows， forehead， nose， and mouth. “About 44 muscles in our face work together to produce 54 types of facial expressions，” says Hu. “Micro- expressions are coordinated， involuntary motions of facial muscles and can change within a 15th of a second. By observing and analyzing these facial features， we can make precise evaluations，” he further explained.

The system was first tested four years ago in a microfinance company for credit risk assessment. According to Hu， previously， the companys 800 offices had to schedule face-to-face meetings before lending. The new technology enabled the company to carry out online evaluations， which resulted in a marked decrease in operating costs and a 40-percent drop on the default rate，” In May 2018， microexpression recognition technology developed by Ping An took first place in the global competition of the One-Minute Gradual Emotion Challenge held in America.

Hu said the loan applicants are first asked two sets of questions. The first set includes questions about their basic information such as age and place of residence， while the second involves queries about their income or outstanding debts. “Then we compare the facial expressions made while answering each set of questions. The results form the basis of our appraisal，” he says.

As with many innovations， the risk for misuse of micro-expression recognition is clear， admitted Hu. Ethical issues arising from the newly developed technology also demand attention. “In theory， this technology could be used in many areas，” he says. “But as far as laws and regulations are concerned， there are narrow limits to its application range. It is also clearly not in our interest for this technology to be abused. We attach great importance to our customers privacy， and our evaluations are all conducted on a voluntary basis.”

The system is currently utilized primarily in Ping Ans quick small loan department. It has played an important role in simplifying application procedures， especially during the COVID-19 pandemic. In a faceto-face conversation， customer service personnel can assess whether the applicant is making honest statements by observing facial expressions.

“Micro-expression recognition analysis is never the sole disqualifier either，” says Hu. “If you fail the initial test， you can still have a chance to get a loan by submitting additional documents and attending a face-to-face meeting with our customer service staff.”