<p>While voice interaction technologies have matured over time, achieving 100% speech recognition accuracy remains challenging in practice. To handle potential misunderstandings and errors, voice robots adopt confirmation strategies to enhance interaction efficiency and accuracy. Previous research primarily focused on confirmation strategies in single-task scenarios, with little attention to multi-tasking scenarios. Through a controlled experiment (N = 65), this research explored user expected confirmation (explicit vs. Implicit) for dialogue task (low vs. High purposiveness) in multi-tasking scenarios. The usability and cognitive load of the confirmation strategies were evaluated. Based on the collected data, users’ expected confirmation strategies in different tasks were analyzed. The results indicate that users experience higher cognitive load when engaged in high-purposiveness dialog tasks compared to low-purposiveness tasks. Explicit confirmation using affirmative sentence structures was found to swiftly capture users’ attention, facilitating error detection and recognition. Users prefer explicit confirmation when performing high-purposiveness tasks to achieve precise task completion. Conversely, when executing low-purposiveness tasks, users prioritize system efficiency and thus lean towards implicit confirmation. This study extends the knowledge boundaries of VUI design by investigating user expected confirmation in multi-tasking scenarios. These findings provide practical implications for developing suitable confirmation strategies for different task types, which are conducive to repairing conversations and improving user experience.</p>