加拿大的税务机构因人力差和AI服务的准确性差而面临批评,专家在扩大AI服务之前要求纠正问题。 Canada's tax agency faces criticism for poor human and AI service accuracy, with experts demanding fixes before expanding AI.

人工智能专家敦促加拿大税务局在扩大人工智能工具之前, 解决其人为客户服务中的持续不准确性, 联邦审计显示, 现场代理人不到五分之一的呼叫者得到正确的答案, 聊天机器人查理只准确约三分之一的时间. AI experts are urging the Canada Revenue Agency to fix persistent inaccuracies in its human customer service before expanding artificial intelligence tools, following a federal audit revealing that fewer than one in five callers received correct answers from live agents and the chatbot Charlie was accurate only about one-third of the time. 专家强调,人工智能依赖于可靠的人类投入,不应取代人类监督,特别是在复杂的税务问题上,主张采用混合模式,聊天机器人处理常规问题,而人类处理细微案件. Experts stress that AI depends on reliable human input and should not replace human oversight, especially for complex tax issues, advocating for a hybrid model where chatbots handle routine questions and humans manage nuanced cases. 他们强调,公众信任要求人工智能的准确性至少达到70%至90%,而人类服务提供的基本改进必须先于人工智能的广泛通过,尽管CRA正在扩大聊天机的操作和基因人工智能的试点测试。 They emphasize that public trust requires AI accuracy of at least 70% to 90%, and that foundational improvements in human service delivery must come before widespread AI adoption, despite the CRA's ongoing expansion of chatbot operations and pilot testing of generative AI.