Interview background:

There is a joke in the industry: Twenty years ago, we focused on how to train people to be as “standard” as machines; twenty years later, we are concerned with how to train machines to be as “human” as people. Therefore, focusing on “machine training” has become the focus of our work at this stage; the introduction of the national occupational definition of “artificial intelligence trainer” has laid the foundation for all of this. And this set of teaching materials and related training system specially customized for the “customer service domain artificial intelligence trainer” closely combined with customer service scenarios will directly affect the commercial implementation level of intelligent agent AI technology.

Column introduction: “Interview with Hospitality Star” is an interview column in the content plan of the “Hospitality Way” streaming media platform (iCustomer.org), created by Customer World Agency. The interviewees are: contestants of the “Future Star” customer center professional skill series competition.

Invite the cutting-edge industry, combined with their own theoretical system and practical experience, to carry out dialogues and interviews on the status quo and future development of the industry.

Reporter: Teacher Meng, hello, I heard from you that your work has nothing to do with artificial intelligence at the beginning, but now you are almost a trainer who can do it alone, and you can share it with our friends What is the opportunity for you to transform from a customer service representative to an artificial intelligence trainer?

Meng Xiaoying: Hello, teacher. In fact, so far, I think I am still a novice in AIT who needs to continue to learn. I am especially grateful to Client World for hosting this competition and giving us such a good platform for learning and communication. I have been in the customer service industry for more than 6 years. The customer service representative is my first professional identity. It can be said that the work experience of the customer service representative has laid a good foundation for my future development. At the beginning of 2020, our China Life Insurance Company headquarters established a project team to build intelligent services. I was lucky to be one of them. This experience became the starting point of my AIT transformation. During the project team, I came into contact with AIT for the first time. Since then, China Life has cultivated the first batch of intelligent trainers, and I am glad that I am among them. Looking back on this journey, every turning point is crucial. I am very grateful for the platform provided by the company and my leaders for their training and support. I want to say: choice and hard work are equally important.

Reporter: I just heard that you mentioned that you started the transformation of your artificial intelligence trainer because of the establishment of a project team at China Life’s headquarters to build intelligent services. We all know that many companies in the industry have been actively embracing digitalization in recent years. With the wave of economic development, for the customer service center, the construction of an intelligent service system is the epitome of customer service in the construction of a digital economy. Could you please briefly introduce to our partners in various enterprises how China Life combines intelligent means in this transformation, Redesign the service system?

Meng Xiaoying: When it comes to redesigning the service system, I can’t help but recall the difficult period when thinking and logic had to be changed. New technology application, business support, service expansion and other needs, the original small number of front-end, one-question-one-answer interactive methods can no longer meet customer needs. When the company’s intelligent customer service products are upgraded, the requirements for service autonomy, convenience, and efficiency have been greatly improved. , From the business level, I understand more about the reshaping of service logic and the refactoring of knowledge system. Why do you say that? The number of our customer service portals has increased by more than 10 times, and the customer coverage is wider. We also put forward higher requirements for the uniformity and differentiation of services, from simply responding to customers and letting customers find answers to directly and accurately answering customers and solving problems The upgrade of service thinking on customer problems means that based on intelligent products and new technologies, more service experience is required and the reconstruction of massive knowledge systems is required. Based on technologies such as knowledge graphs, central control centers, and multi-round dialogues, we have built hundreds of multi-round dialogue streams and over a thousand product knowledge graphs based on high-frequency customer questions, business logic, and service experience. In the case of 1 times more, the knowledge base can be reduced by about 40%, which is more in line with the company’s business development and service supply needs.

Reporter: Thank you, Mr. Meng, for sharing. Being able to see the experience at that time is very important for the intelligent transformation of you and the enterprise. Next question, this time we hold the Future Star Trainer Competition to “promote learning through competition and promote through practice”. For the training of intelligent services, many small partners are most troubled by the synergy between artificial intelligence and intelligence. Should it be diversion or satisfaction? Please also share with your friends, how do you control the relationship between intelligent services and manual services?

Meng Xiaoying: I think the relationship between the two has a lot to do with the service concept of an enterprise. In this era full of smart products, it is easy for us to enjoy smart services, but sometimes there will be “finding Human Difficulty” situation. In terms of customer service in our company, what we pursue is to provide customers with “simple, high-quality, and warm” services. Obviously, manual service is a very important part of it, so it is very easy to find manual labor in our company’s service channels For the time being, warm service and emotional resonance between people cannot be replaced by smart services.

Reporter: Thank you, Teacher Meng. In other words, intelligence cannot completely replace artificial intelligence, it is only one part of the entire service link. Next, please share with us about the training and operation of smart service products. What kind of method should our friends open correctly?

Meng Xiaoying: From the perspective of time, the training and operation of smart services can be divided into regular training and irregular training. Regular training is a routine daily/weekly quality inspection service log. It is determined whether training is needed based on the quality inspection results. The training mainly depends on the company’s business development, and it is decided whether to train according to the actual business needs and customer needs.

Reporter: Thank you very much for Mr. Meng’s summary. At present, our competition has entered the semi-final stage. At this stage, we mainly assess the management ability of the original corpus data of the contestants for smart service products, including: the collection, cleaning and labeling of the original corpus Wait, can you please share your knowledge in this area, and also give us some inspiration and ideas for participating friends.

Meng Xiaoying: I think most of the partners who have experienced the construction of intelligent customer service have felt the importance of real service logs. The original data of intelligent services is undoubtedly an invaluable treasure for our artificial intelligence trainers, so the management and application of these data is also very important . Whether the smart service is good or not depends on the evaluation of customers. In addition to customer satisfaction, the original service log is also a good way to reflect. The service demand of smart service comes from customers, and it will naturally benefit from customers. First of all, the collection and cleaning of service data is the key, and more people can be involved in it, such as daily quality inspection, sharing of other service channels, etc., to obtain as much high-quality original data as possible, and then it is possible to get more Excavate the real needs of customers, so as to adjust our knowledge system to better meet customer needs. Secondly, the labeling of service data is the most important part of training. A mature service knowledge system contains knowledge of multiple sections, because each person’s understanding of semantics may be biased, and the same person’s labeling results are the best. Larger, then the division of labor can be carried out according to the knowledge section, which can reduce the labeling deviation and improve the training effect.

Reporter: Okay, thank you very much, Mr. Meng. So finally, I would like to ask you a question on behalf of our fellow trainers: The end point of smart service products is to interact with customers and respond to customer questions, so how should we improve the response effect?

Meng Xiaoying: The key to improving the response effect is first of all to grasp the position. Our service objects are customers, so we must stand in the perspective of customers and readers to design answers, treat customers as professional novices, and use more altruistic thinking , The answer writing should be easy to understand, so that readers will find it useful, clear and beneficial after reading it.

Reporter: Okay, thank you very much for Teacher Meng Xiaoying of the Mongolian Horse Team for being a guest in our interview with Hospitality Stars in this issue. He brought our little friends your new journey from customer service representative to artificial intelligence trainer at that time. Thank you again for sharing your training ideas and knowledge from the perspective of competition and actual training operations.

Reporter: Liu Zixin

Written by: Qian Yi

Planning: Su Yu

Time: June 2022

Location: Online

Team introduction: The team members (Meng Xiaoying, Ma Wenlong, Li Dandan) are all from Inner Mongolia Branch of China Life Insurance Co., Ltd. The team members have participated in the whole process of intelligent customer service construction, and currently operate the online robot independently. Rich customer service experience is their confidence, and they cherish every experience, after all, there is no way to go in vain in life. ?