I wonder if you have ever had such an experience? When you call the customer service hotline or click on the customer service function of a website, an "AI customer service" will always appear. During the conversation, the "AI customer service" cannot understand the words and cannot solve the problem. When you call "switch to human service" over and over again, what you are waiting for is "Sorry, the human agents are currently busy"...

The original intention of the artificial intelligence customer service system is to replace part of the manual customer service to understand user needs, answer relevant questions, and provide solutions. However, why do artificial intelligence customer service frequently destroy the user experience "unintelligently" and even persuade users to quit?

AI customer service becomes a communication “wall”

Impact on user experience

When the content they want to consult is not among the regular options listed by AI customer service, users often hope to be transferred to manual services. However, at this time, AI customer service seems to be like a "talking wall", blocking the channel for effective communication between users and the platform.

Ms. Ma, a consumer, has had the experience of being "dissuaded" by AI customer service many times. What impressed her deeply was a broadband transfer business. After Ms. Ma dialed the operator's customer service number, she could not find the question she wanted to consult among the options provided by the AI ​​customer service. She could only apply for manual transfer, and it took 4 transfers before finally connecting to manual service.


Although these AI customer service on various platforms can to a certain extent replace manual answers to some preset high-frequency questions, there are still a large number of user reports that in some scenarios, AI customer service has problems such as answering questions incorrectly, talking to itself, and being too templated.

Reporters have recently tested the customer service systems of more than 10 mainstream e-commerce, social, financial, logistics and other platforms and found that whether it is online customer service or telephone customer service, AI customer service has obvious deficiencies in understanding, while manual customer service has many barriers to access, seriously affecting user experience.


AI customer service claims to reduce costs at low prices

In fact, the level of quality varies greatly.

Although there are many problems with AI customer service, many companies and stores still choose to purchase AI customer service systems. On various social media, there are many AI customer service providers claiming to be low-priced and easy to use.

Mr. Fu, a salesperson at a service provider, told reporters that the first-month trial of the AI ​​customer service product only costs 299 yuan, and the annual service package is 6,999 yuan. Calculated based on the same workload, it is much more cost-effective than hiring manual customer service. "It is automatically learning. You only need to 'feed' it the vocabulary the store needs. If you have no special requirements, we also have our own industry vocabulary package."


Some basic AI customer service system products can only handle simple inquiries, and tend to have a "template" feeling when replying to users. In order to increase the "IQ" of AI customer service and make it more humane, additional costs need to be spent on enriching and "feeding" the knowledge base.

Mr. Li, a technician from a technology company, said that a robot is like a new employee. You need to train it. "The trainer will tune the robot every year, and the cost is 5,000 yuan a year. If it is not tuned, the robot will be 'stupid'."

Enterprises set up a "reply to the bottom" strategy

Manufacturing transfer manual difficulty

Professionals said that the current problem that consumers generally report about the difficulty of manual transfer is actually a technical means. Technicians from a technology company that specializes in providing online customer service systems, intelligent customer service robots and other products to enterprises told reporters that companies can independently decide how easy it is for users to "find" human customer service, and technicians can set up a "cover-in reply" strategy for the AI ​​customer service system based on the needs of the company.


Mr. Li explained that in order to reduce labor costs, many companies will proactively set up obstacles to jump to manual agents. Since AI customer service cannot fundamentally solve user problems at this stage, some of its functions will be adjusted to "intercept user requests."

"For example, after a few rounds of dialogue, the process will be transferred to manual work, or if the customer is in a hurry, the communication can be transferred directly to manual work. If the problem can be solved with AI, it is generally not allowed to be transferred to manual work. Doesn't switching to manual work cost the money of human customer service?"

This kind of "service" idea is more prominent among small merchants. Some e-commerce practitioners revealed on social media that the platform has strict requirements for merchants' response time, and points will be deducted for failure to respond within the specified time. Considering that manual customer service has high work intensity and high mobility, recruitment and training are costly. Regardless of whether AI customer service is intelligent enough or not, it has to use a more cost-effective one to "deal with" customers.

Excessive pursuit of “cost reduction and efficiency increase”

Causing AI customer service to become “aphasic”

Regarding the various non-"intelligent" problems currently existing in AI customer service systems, industry experts believe that the current value orientation deviations and technical deficiencies of some enterprises and technical service providers have made the "aphasia" of AI customer service services a common phenomenon.

Major General Ding, an IT industry expert, said that many companies are now excessively reducing costs and increasing efficiency, defining customer service as a cost department, not investing enough in technology, and using "interception rate" instead of "resolution rate" as KPI, forming a vicious circle.

Mr. Yu, a practitioner in the artificial intelligence industry, observed that in the perception of many companies or merchants, these "unintelligent" AI customer service prompts customers - "Your problem is not important, so don't come to me." But in Mr. Yu's view, they are using an "expensive low-cost tool" that comes at the expense of customer disappointment.

Artificial intelligence practitioners said that from a technical perspective, for companies or merchants using AI customer service systems, the future breakthrough point still lies in "human-machine collaboration", that is, through continuous optimization, AI customer service can understand customers. On the other hand, in terms of user experience, merchants should also consider the needs of consumers more. At this stage, AI customer service lacks the sincerity to solve problems and lacks the ability to perceive emotions that humans have. This is not only a lack of technology, but also a lack of "warmth" of consumer service by the businesses behind the technology.