According to a senior official at the Canada Revenue Agency (CRA), the agency is aiming to enhance the accuracy of responses provided by call centre agents by leveraging artificial intelligence and improved training methods.
Melanie Serjak, an assistant commissioner at the CRA, informed members of a standing committee that the agency plans to implement a more standardized and advanced level of training while incorporating automation to ensure that CRA agents offer precise guidance to the public.
Speaking to MPs, Serjak mentioned, “We are exploring the use of artificial intelligence and other technological tools to support agents in delivering comprehensive and precise answers. We are presently collaborating with our vendor in this area.”
An audit conducted by Auditor General Karen Hogan’s office revealed that only 17% of individual tax inquiries were accurately addressed by CRA call centre staff after making calls to the agency’s contact centres over a four-month period this year.
In a recent report, Hogan criticized the CRA for prioritizing employee schedules over the accuracy and completeness of information provided to callers.
Auditor General Karen Hogan has found that Canada Revenue Agency contact centres are repeatedly failing to answer calls in a timely manner — and when agents do connect to a customer, they often provide inaccurate responses.
Hogan emphasized that the CRA has numerous opportunities to enhance its performance through training and call triaging.
Serjak explained that agents undergo between two to 13 weeks of classroom training and additional live training alongside senior agents before handling calls.
“Regarding training, we aim to introduce a more advanced and standardized level of trainer,” she stated. “We are also planning to implement automation in our quality review process soon to enhance efficiency and effectiveness for our quality evaluators.”
The CRA has employed a virtual chatbot named Charlie to provide automated responses to commonly asked questions.
Hogan’s report highlighted that the chatbot Charlie provided more accurate responses (33%) compared to human agents when asked about personal tax matters, indicating room for improvement in agent performance.
“Charlie’s accuracy rate of 33% surpasses that of speaking with an agent directly about personal tax inquiries,” Hogan stated. “This underscores the need for substantial improvement.”
The audit found that only nine percent of agents’ performance evaluation score related to the accuracy and completeness of information provided to callers, while 45% was linked to schedule adherence and call handling time.
“The minimal focus on accuracy fails to prioritize