Paper Type
Short
Paper Number
PACIS2026-1130
Description
Multitasking is common in workplaces, but its implications may change with the growing integration of agentic AI. This study examines how human multitasking affects service performance in hybrid service systems where human agents collaborate with AI partners. We collaborate with a leading Chinese smart-home company that has adopted a hybrid customer service system and analyze a granular dataset of session-level chat transcripts and operational records from May to July 2024. To address endogeneity in multitasking, we employ a control function approach with an instrumental variable based on short-term shocks in peers’ session completions. Our findings reveal that one- unit increase in multitasking increases response delay by 1.706 seconds and reduces customer emotion by approximately 1.14%. Drawing on the IS delegation framework, we show that multitasking disrupts effective human–AI task delegation, increasing delegation misalignment, which in turn amplifies response delays and worsens customer emotion. Further mediation analysis validates our theory
Recommended Citation
Chen, Junhang; Yang, Cenying; and Fu, Xin, "When Human Multitasking Meets Agentic AI" (2026). PACIS 2026 Proceedings. 1.
https://aisel.aisnet.org/pacis2026/ai_fow/ai_fow/1
When Human Multitasking Meets Agentic AI
Multitasking is common in workplaces, but its implications may change with the growing integration of agentic AI. This study examines how human multitasking affects service performance in hybrid service systems where human agents collaborate with AI partners. We collaborate with a leading Chinese smart-home company that has adopted a hybrid customer service system and analyze a granular dataset of session-level chat transcripts and operational records from May to July 2024. To address endogeneity in multitasking, we employ a control function approach with an instrumental variable based on short-term shocks in peers’ session completions. Our findings reveal that one- unit increase in multitasking increases response delay by 1.706 seconds and reduces customer emotion by approximately 1.14%. Drawing on the IS delegation framework, we show that multitasking disrupts effective human–AI task delegation, increasing delegation misalignment, which in turn amplifies response delays and worsens customer emotion. Further mediation analysis validates our theory
Comments
02-FutureofWork