智能客服长期因技术不成熟而饱受诟病，但这一局面正在改变。随着大模型应用的推进，AI客服开始具备理解上下文、调用工具、处理复杂任务的能力，从单纯的问答系统向完整的服务流程升级[1]。不过用户的不满情绪仍在高涨——2024年全国消费者投诉中智能客服相关投诉达6969件，同比增长56.3%[1]。
当前市场格局中，大型科技公司与垂直创业企业并行发展。根据IDC数据，2024年智能客服五大主要厂商市场份额达35%，其中阿里云占11.4%、百度智能云占10.4%[1]。商业模式也在探索创新，部分企业开始从传统SaaS订阅制向按结果付费转变。国外客服服务商Sierra通过这一模式完成9.5亿美元融资，估值超150亿美元[1]。
技术进步体现在更精细的用户体验设计上。中国电信推出的GOAT-SLM大模型声称能识别用户语气中的焦虑，并生成具有共情意义的回应[1]。然而，AI客服在复杂售后、风险判断等场景仍存在瓶颈，需要人工介入[1]。浙江省消保委建议，常规咨询可由智能客服处理，但投诉及涉及资金安全的问题应优先转接人工客服[1]。
Intelligent customer service has faced persistent complaints over technical limitations for years, but emerging large language model applications are transforming the sector [1]. AI customer service systems are now developing capabilities to understand context, invoke tools, and manage complex service workflows, moving beyond simple question-and-answer interactions toward comprehensive transaction processing [1].
The improvement comes amid growing market adoption. According to the State Administration for Market Regulation, consumer complaints related to intelligent customer service reached 6,969 cases in 2024, representing a 56.3% year-on-year increase [1]. Despite the rising complaints, IDC data shows the top five intelligent customer service providers captured 35% market share in 2024, with Alibaba Cloud holding 11.4% and Baidu Intelligent Cloud at 10.4% [1]. Major enterprises and vertical startups are developing in parallel, with Sierra completing a $950 million funding round at a valuation exceeding $150 billion and adopting a results-based payment model [1].
Technology innovations are enabling more sophisticated interactions. China Telecom launched its GOAT-SLM large model, claiming the ability to detect anxiety in users' tone and generate empathetic responses [1]. However, the industry recognizes current limitations. The Zhejiang Consumer Protection Committee recommends that routine inquiries be handled by intelligent systems, while complaints and finance-related matters should be prioritized for human agent transfer [1]. Complex post-sale scenarios and risk assessments continue to require human intervention, indicating that AI customer service remains complementary rather than fully replacing human support [1].