TL;DR:
- Chinese scholar introduces Tong Tong, the world’s first AI child, at a tech exhibition in Beijing.
- Tong Tong, created at BIGAI, showcases advanced cognitive AI capabilities, distinguishing right from wrong and shaping her own tasks.
- BIGAI’s “Tong Test” assesses AI models across multiple dimensions, bringing AGI closer to reality.
- Zhu Songchun emphasizes practical skills and values in AGI development.
- Despite skepticism, Tong Tong represents a significant step toward achieving general artificial intelligence.
Main AI News:
A groundbreaking revelation came forth from a Chinese scholar at the recent Frontiers of General Artificial Intelligence Technology Exhibition in Beijing. In what is being hailed as a monumental stride in the field of artificial intelligence, visitors had the unique opportunity to interact with an AI creation known as Tong Tong, a virtual toddler affectionately referred to as “Little Girl” in English.
The mastermind behind Tong Tong is Zhu Songchun, a computer scientist and founder of the Beijing Institute for General Artificial Intelligence (BIGAI). Zhu’s expertise lies in the realm of “cognitive artificial intelligence,” a field dedicated to replicating human cognitive abilities using AI technology.
What sets Tong Tong apart from other AI avatars is her remarkable ability to not only perform tasks assigned to her within a virtual environment but also to autonomously generate new tasks for herself. In a translated statement from BIGAI’s Chinese-language video, it is emphasized that Tong Tong possesses a mind of her own, distinguishing right from wrong, expressing her viewpoints in various scenarios, and holding the potential to shape the future.
Beyond serving as a captivating spectacle for conference attendees, Tong Tong is poised to play a pivotal role in the advancement of Artificial General Intelligence (AGI) through BIGAI’s visionary “Tong Test.” Proposed as a more intricate counterpart to the Turing Test, this extensive assessment comprises nearly 150 questions, aiming to gauge the sophistication of an AI model. The evaluation delves into five measurable dimensions: vision, language, cognition, emotion, and learning, along with an assessment of social and group values. Its primary objective is to determine the proximity of a given AI model to achieving AGI.
Zhu Songchun elucidated the significance of the Tong Test, stating, “For [AGI] to seamlessly integrate into human environments, it must demonstrate the capability to learn and execute tasks in complex settings, guided by values and a profound understanding of causality. This is the driving force behind the Tong Test—a novel approach to evaluating general AI, emphasizing practical skills and values.”
While this ambition may bear semblance to the AI optimism often met with skepticism in the United States, where Zhu spent a significant portion of his life before returning to China, Tong Tong embodies Zhu’s relentless pursuit of AGI. As he remarked during the Beijing conference, the data gathered through Tong Tong’s interactions may propel humanity closer to the formidable goal of achieving general artificial intelligence. Zhu asserted, “To advance toward the realm of general artificial intelligence, we must create entities capable of comprehending the real world and possessing a diverse array of skills.“
Conclusion:
The introduction of Tong Tong, the AI-powered simulated child, highlights China’s strides in AI technology. BIGAI’s innovative “Tong Test” signifies a major leap toward achieving Artificial General Intelligence, with a focus on practical abilities and values. This development underscores the growing importance of AGI in the global market, offering potential opportunities and challenges for businesses in the AI sector.
