<p>With the rapid advancement of artificial intelligence (AI) technology, AI-driven virtual streamers have become a new live streaming selling mode for many brands. This study explores how brands can balance the advantages and disadvantages of virtual and human streamers to proactively adapt to new live streaming commerce modes. To explore whether and how brands should introduce virtual streamers, we develop live streaming optimization models for brands under three selling modes: independent selling by human streamers (HN), collaborative selling between human and virtual streamers (HA), and independent selling by virtual streamers (AA). Furthermore, we establish two extended models considering the impact of consumer switching and waiting costs, as well as absence risks faced by human streamers on the brand and consumers. The results show that whether and how to introduce virtual streamers depends on the degree of consumer distrust in virtual streamers and the technological development barrier coefficient. When both these factors exceed a certain threshold, brands should adopt the HN mode instead of introducing the AA mode. Conversely, when these factors are below a certain threshold, brands can adopt AA mode. If the factors fall between these thresholds, the HA mode yields higher profits. Moreover, social welfare is higher in the HA mode than in the HN or AA modes.</p>