中国消费者协会统计显示,今年上半年AI客服投诉成为新热点,AI客服答非所问、“已读乱回”引发消费纠纷1。在实际案例中,有中国民众轻信AI关于机票退改签的建议,实际被扣手续费高达40%,损失人民币数百元1。中消协投诉部副主任郭耀宗指出,消费者日常接触的人工智能客服主要分为两类:通用生成式人工智能服务和经营者的人工智能客服1。才博智慧治理研究院院长吴岩松表示,只要是在企业App或官方微信、微博、小程序这些平台上回复,都应该代表公司的立场1。专家同时指出,企业需对智慧客服的回复内容负责,不能以算法生成内容不具法律效力为由拒绝履行承诺1。
为规范人工与AI客服协同流程,中国首个人工客服与智慧客服协同机制的国家标准《顾客联络服务,人工与智慧客户服务协同要求》将于9月1日起实施1。该标准规定,智慧客服优先处理标准化问题,人工客服优先处理复杂、涉安全等问题1。同时,二者之间切换需同步客户信息且不重复询问1。
Complaints regarding artificial intelligence customer service have become a new hotspot in consumer disputes handled by the China Consumers Association in the first half of this year 1. The issues primarily involve AI systems providing irrelevant answers or replying randomly after reading messages, which has led to consumer disputes 1. For instance, a Chinese consumer followed an AI's advice on flight ticket refunds and changes, only to be charged a handling fee of up to 40%, resulting in a loss of several hundred RMB 1.
To address these issues and regulate the coordination between human and AI agents, China's first national standard for customer contact services, titled "Customer Contact Services - Requirements for the Coordination of Human and Intelligent Customer Services," will take effect on September 1 1. Under the new standard, intelligent customer service will prioritize handling standardized issues, while human agents will prioritize complex and safety-related problems 1. Furthermore, transitioning between the two systems requires synchronizing customer information to avoid repetitive questioning 1.
Guo Yaozong, deputy director of the complaint department at the China Consumers Association, pointed out that the AI customer services consumers encounter daily are mainly divided into two categories: general generative AI services and operators' AI customer services 1. Wu Yansong, president of the Caibo Smart Governance Research Institute, stated that as long as responses are provided on corporate platforms such as official apps, WeChat, Weibo, or mini-programs, they should represent the company's stance 1. Experts pointed out that enterprises must take responsibility for the replies generated by their intelligent customer service and cannot refuse to fulfill commitments by claiming that algorithm-generated content lacks legal validity 1.
评论
还没有评论,欢迎留下第一条。