Do Virtual Stars Really Need a Human Touch?

Virtual assistants are popping up everywhere, especially in ads. But how do we decide if we can trust them? A recent study looked at how their looks and voices influence our trust, especially when these features are created by AI. Researchers picked a well-known TV character, Sheldon Cooper, and made different versions of him using AI tools. Some versions had hand-made details, others used AI to rebuild his features, and some were just plain digital copies. His voice was also tweaked—either cloned from real speech or fully synthetic.The experiment showed that people trusted the virtual Sheldon more when he looked more realistic or familiar. His voice mattered too, but only in how real or recognizable he felt. Even the best AI still can’t fully match real video footage. The study also found that trust doesn’t always grow in a straight line—sometimes, small changes in how close Sheldon "stood" to people affected trust in unexpected ways. This raises questions about how we interact with digital characters. If AI keeps improving, will we ever fully trust virtual celebrities? Or will something always feel off?