TikTok parent ByteDance has upgraded its Doubao chatbot, one of China’s most popular consumer-facing artificial intelligence (AI) apps, with the recent launch of a real-time video call function that turns the app into an interactive digital assistant.
The function allows users to engage in an interactive video conversation with the AI technology behind the chatbot, according to an announcement last Friday via Doubao’s WeChat account. Users can activate the function by simply turning on their smartphone’s camera during a voice call.
Once activated, Doubao can automatically serve in real time as a docent during a museum tour; a tutor with gardening knowledge when observing plants; a recipe master when shopping for ingredients at a grocery store; and an analyst when studying charts, graphs or videos.
According to Doubao, the new function was built on ByteDance’s visual reasoning AI model, which integrates visual and language inputs to support content creation and in-depth study of subject matter. It also supports online search to obtain the latest information from the internet.
Doubao’s real-time, interactive video call function shows the latest progress made by ByteDance in generative AI (GenAI), showcasing the advanced multimodal capabilities of products built on its own AI models. GenAI refers to algorithms used to create new content, including audio, code, images, text, simulations and videos.
Earlier this month, Doubao promoted its capabilities to turn any photo into pixel art. In February, parent ByteDance introduced its OmniHuman-1 multimodal AI model, which gained widespread attention for its ability to transform photos and sound bites into realistic videos.
