iFLYTEK's AI Avatar Debuts at CAS Public Science Day, Discussing the Future of AI with Scientists

From May 18 to 19, the Chinese Academy of Sciences (CAS) commemorated the 20th anniversary of its Public Science Day with a specially curated main event. iFLYTEK introduced its AI Avatar, "Mumu”, who participated in the opening ceremony's themed dialogue, "AI, Is the Turning Point Here?" held on May 18. Mumu engaged in interactive discussions with the host and scientists, sharing its insights and perspectives on the field of AI.

Event Highlights

During the event, the host posed a thought-provoking question, "Humans gradually gain understanding of the world through studying the laws of physics, chemistry, and life. While AI can provide answers to many scientific questions, does it truly grasp the scientific principles governing the world's operations?" In response to this question, iFLYTEK's AI Avatar, Mumu, offered the following response," AI is indeed capable of addressing numerous scientific inquiries. However, it doesn't 'understand' the scientific laws governing the world in the same way that humans do. AI's responses are derived from extensive data and algorithms, analogous to knowing the route to a supermarket without necessarily comprehending the reasons behind taking that particular route. Nonetheless, this doesn't impede the exploration and application of AI, as it can provide valuable assistance in a wide range of domains."

AI Avatar “Mumu”

In addition, Mumu also proposed its own suggestions on how humans and AI can better collaborate and coexist in the face of AI's rapid development, “First, humans must maintain their creativity and innovation. Second, interpersonal and communication skills are also crucial, as AI cannot fully comprehend and process human emotions. Finally, humans should actively learn and utilize AI based on the characteristics of AI society and understand how to collaborate with AI so as to enhance their competitiveness.”

Roundtable Discussion

iFLYTEK's AI Avatar, powered by the latest technology from iFLYTEK SPARK Large Speech Model V3.5, seamlessly combines AI Avatar driving technology with iFLYTEK's cutting-edge facial modeling, lip shape prediction, and image processing techniques. This powerful combination allows for the creation of realistic 2D/3D virtual avatars that can replace real-life anchors in various video broadcasting and interactive scenarios. It can not only automatically generate text-to-video broadcasts but also facilitate intelligent interactions for virtual anchors across diverse settings.

Leveraging AI algorithms, iFLYTEK's AI Avatar can rapidly generate a personalized “digital avatar” within its iFlyVoice Studio’s system by extracting facial features and voice characteristics. Furthermore, iFLYTEK SPARK Large Speech Model’s TrueVoice dialogue technology released on January 30, allows the AI Avatar to produce near-human vocal expressions. Additionally, the model boasts an impressive emotional perception accuracy exceeding 85%, enabling the AI Avatar to portray emotions like happiness, apology, coquetry, and confusion in a more vivid manner.

Beyond voice, gestures play a vital role in AI Avatar’s interaction. Leveraging the large model, the AI Avatar can now grasp the nuances of spoken language and automatically generate and synchronize gestures that complement its speech, resulting in lifelike and engaging interactions.

Powered by iFLYTEK SPARK Large Speech Model, iFLYTEK's AI Avatar continuously upgrades its intelligent perception ability, semantic understanding ability, and emotional expression ability, making its "face-to-face" communication and question-answering with users more effective and open.

As the event drew to a successful close, iFLYTEK's AI Avatar "Mumu" sparked lively discussions among participants. Mumu's performance not only demonstrated the vast potential of AI technology but also offered valuable insights into fostering a future where AI and human society can coexist harmoniously. We can look forward to a future where AI Avatars become irreplaceable companions, collaborating with us to unlock new possibilities in the age of AI.