The aim of this study is to evaluate the effects of a teaching simulation activity that uses a chatbot on preservice teachers' efficacy. Forty-six preservice teachers were asked to teach the chatbot the topic of school violence and how to handle it. They were assigned to one of three groups: Teaching a chatbot whose attitude was impolite, polite, or ordinary. The participants completed a teacher efficacy test at the pretest and posttest. The results show that the participants who taught the ordinary chatbot significantly increased their teacher efficacy levels. However, an Analysis of Covariance shows that the posttest scores of student engagement were not significantly different due to the group setting. The data of participants' conversations with the chatbots and the participant interview data revealed that the ordinary group had more opportunities to increase their mastery experiences than the other groups did. It can be suggested that designing virtual students with ordinary and regular attitudes and behaviors seems appropriate to provide preservice teachers with teaching opportunities to increase their teaching efficacy levels.