A Meta-Analysis on the Effects of Using Artificial Intelligence–Powered Virtual Agents in Simulation-Based Learning

• In Event: Virtual Agents, Artificial Intelligence, and Automation

Abstract

This meta-analysis examined the effects of AI-powered virtual agents in simulation-based learning in response to a gap in the literature. Forty-nine effect sizes derived from 22 studies suggested a medium positive overall effect, g ̅ = 0.43, SE = 0.08, 95% C.I. [0.27, 0.59]. Further, moderator analyses revealed that intervention length and AI technologies significantly explain the heterogeneous effects; whereas It’s universally effective across the role-specific functionality of AI-powered virtual agents. Overall, this meta-analysis provides systematic evidence and promising prospects for the adoption of AI-powered virtual agents in simulation-based learning. Educational researchers and practitioners can refer to our findings to make evidence-based design decisions on intervention length, the role-specific functionality, and AI technologies when using AI-powered virtual agents simulation-based learning.  

Authors

• Chih-Pu Dai, University of Hawai‘i at Mānoa
  Presenting Author

• Fengfeng Ke, Florida State University
  Non-Presenting Author

• Yanjun Pan, Southern Methodist University
  Non-Presenting Author

• Jewoong Moon, University of Alabama
  Non-Presenting Author

• Zhichun Liu, University of Hong Kong
  Non-Presenting Author