She can distinguish between right and wrong, express her views in various situations, and has the ability to shape the future.
A Chinese academic has introduced what he claims to be the world’s first AI child, suggesting that this creation could usher in a new era for technology.
According to the South China Morning Post, attendees at the Frontiers of General Artificial Intelligence Technology Exhibition in Beijing in late January had the opportunity to interact with an avatar representing Tong Tong, a virtual toddler named “Little Girl” in English.
Developed at the Beijing Institute for General Artificial Intelligence (BIGAI), which focuses on constructing artificial general intelligence (AGI), Tong Tong is the brainchild of Zhu Songchun, the institute’s founder and a computer scientist specializing in cognitive artificial intelligence, aimed at emulating human cognition.
While AI avatars can vary in appearance and personality, Tong Tong is intended to break new technical ground by not only performing given tasks in a virtual environment but also autonomously creating new tasks for herself.
According to a Chinese-language video from BIGAI translated by SCMP, “Tong Tong has a mind of her own and seeks to comprehend the common sense taught by humans. She can distinguish right from wrong, express her attitudes in various situations, and has the ability to shape the future.”
Logical Deduction: Beyond serving as a novelty for conference attendees, Tong Tong is expected to have a significant role in the future of AGI through BIGAI’s “Tong Test.”
Introduced last year as a complex assessment similar to an inverse Turing Test, this nearly 150-question evaluation aims to gauge the sophistication of an AI model. It assesses five quantifiable dimensions—vision, language, cognition, emotion, and learning—as well as social and group values to determine the AI model’s proximity to AGI.
Zhu stated in a previous statement quoted by SCMP, “For [AGI] to seamlessly integrate into human environments, it must learn and execute tasks in complex settings, driven by values and an understanding of causality. This is why we proposed the Tong Test, a new direction for testing general AI, focusing on practical abilities and values.”
While this may seem overly optimistic about AI, Tong Tong is an integral part of Zhu’s endeavors to develop AGI. As Zhu mentioned at the Beijing conference last month, the data she contributes to collecting may bring humanity closer to that daunting objective.
“To progress towards general artificial intelligence,” Zhu emphasized, “we must create entities capable of understanding the real world and possessing a broad range of skills.”
