A 200-square-meter live-streaming studio stands completely empty, yet dozens of computers display hosts selling products. Such short videos flooded social media platforms, with captions like “All AI digital live streams” striking a chord with the public. These “hyper-realistic 2D digital avatars” displayed natural expressions and smooth delivery—without the empty seats giving them away, it would be hard to distinguish virtual from reality. Fueled by ChatGPT, “unmanned live streaming studios” have quietly emerged. Industry insiders reveal that digital hosts operate for three days without glitches, achieving 15% higher conversion rates than newcomers. Cost advantages are even more pronounced: human hosts cost around 30,000 yuan monthly with risk of broadcast interruptions; digital hosts require a one-time investment of roughly 50,000 yuan and only 2,000 yuan in monthly maintenance, enabling 24/7 streaming. One beauty company deployed ten digital hosts managed by just two operators, boosting sales by 40%. However, cloning real people risks portrait rights disputes—some competitors have already paid hundreds of thousands in compensation.
AI engineers reveal that digital humans are modeled to generate appearances, brought to life through motion capture and voice synthesis. Inputting keywords can generate three avatars in three minutes; importing scripts enables product explanations, with tone adjustments based on merchandise. After integrating large language models, they can recognize and respond to real-time comments with over 90% accuracy, even switching tones based on comment sentiment to mitigate risks. New hosts worry about being “replaced,” but digital humans currently only take over standardized roles like explaining beauty products or snacks. For experiential categories like apparel and luxury goods, digital human penetration remains below 5% because they cannot convey authentic experiences or tell design stories. Industry association experts predict a future “human host + digital assistant” model, where the former delivers core content while the latter handles basic inquiries.
Concerns follow: Consumers have complained about false claims like “masks reduce nasolabial folds,” making accountability difficult. “AI face-swap livestreaming” services generate lifelike digital avatars from photos alone, posing fraud risks. Multiple regions now mandate “virtual host” labels for such streams, requiring companies and platforms to establish verification systems. Digital avatars have now expanded into government services and training sectors. Industry insiders predict they may become standard in live streaming within 3-5 years. Yet the consensus remains: “Trust is the core of live streaming. Digital avatars boost efficiency, but retaining viewers still relies on authentic content and reliable products.” Behind the virtual host's smile, this transformation compels practitioners to ponder: Amidst the technological wave, where does humanity's irreplaceability lie?