China has introduced the world’s first AI child TongTong 2.0, an upgraded artificial general intelligence (AGI) system with remarkable human-like abilities. Unveiled at the 2025 Zhongguancun Forum in Beijing, this cutting-edge AI model is designed to think, learn, and even express emotions similar to a 5- to 6-year-old child, surpassing the cognitive abilities of its previous version.
Developed by the Beijing Institute for General Artificial Intelligence, TongTong 2.0 represents a leap forward in AI research, showcasing the potential of open-source development in accelerating AGI advancements. This innovation further cements China’s position as a leader in next-generation AI technologies.
TongTong’s Evolution: From Toddler to Negotiator
Originally launched in early 2024, TongTong 1.0 functioned at the intelligence level of a 3-year-old. However, TongTong 2.0 has now evolved significantly, improving in language processing, cognition, movement, learning, emotions, and interaction. The AI child can now negotiate, delay tasks, and even throw tantrums, much like a real child.
One of its most fascinating features is its ability to engage in value-driven behavior. In simulated environments, TongTong 2.0 demonstrates an understanding of cleanliness, autonomously tidying up toys, wiping surfaces, and organizing objects. These actions stem from advanced AI training conducted in over 100 hyper-realistic digital environments, enabling it to develop knowledge, social behaviors, and decision-making skills.
A Digital Persona with a Mind of Its Own
According to Chen Hao, Executive Director of the Advanced Technology Center at the Beijing Institute for General Artificial Intelligence, TongTong 2.0 has its own set of values and worldview. This means the AI doesn’t just respond to commands; it processes intent, considers its own values, and engages in contextually aware conversations.
For instance, if asked to go to bed or eat something it dislikes, TongTong 2.0 might negotiate, justify its actions, or even request a reward before following instructions—just like a real 6-year-old child. This level of conversational coherence and personality-driven responses marks a significant milestone in AGI development.
Adapting to Real-World Situations
Unlike its predecessor, which struggled with unpredictable environmental changes, TongTong 2.0 can now reflect, re-plan, and adapt in real-time. It can pause, modify, and even cancel tasks as needed, dynamically adjusting its approach to meet its goals more efficiently.
One of its most impressive capabilities is autonomous learning. Rather than relying solely on preloaded data, TongTong 2.0 interacts with humans, processes real-world experiences, and updates its knowledge base in real time. This human-like introspection enables it to refine its values, skills, and cognitive abilities continuously.
Problem-Solving Like a Child
During a demonstration, engineers challenged TongTong 2.0 by placing a remote control out of reach. Instead of giving up, the AI child removed its shoes, used sofa cushions to gain height, and successfully retrieved the device—mimicking a problem-solving strategy typical of young children.
Another example showcased TongTong’s growing intelligence. When the AI child failed to turn on an air conditioner, it learned through trial and error that the device needed to be plugged in. After resolving the issue, it stored the knowledge for future use, demonstrating real-world learning and adaptability.
China’s Race to AGI Leadership
The unveiling of TongTong 2.0 is a testament to China’s rapid progress in AGI and its commitment to open-source AI innovation. By allowing AI models like TongTong to evolve and interact with real-world environments, China is pushing the boundaries of human-like artificial intelligence.
As AGI research accelerates, AI models like TongTong 2.0 may pave the way for fully autonomous AI assistants, lifelike digital personas, and even AI-powered education tools. With its ability to learn, reason, and exhibit emotions, TongTong 2.0 brings us one step closer to AI systems that behave and think like humans.
While AI ethics and regulations remain a crucial consideration, there’s no doubt that the world’s first AI child is a groundbreaking milestone in artificial intelligence history.
Get the Latest AI News on AI Content Minds Blog