Chinese Scientists Develop AI `Child´ with `Human Emotions´

    Published on:

    Chinese scientists have developed an AI known as Tongtong that they claim can “interpret human emotions.” Bigai calls his new creation an “AI child,” but in reality, it's just software trained to perform certain tasks on its own without human intervention.

    Bigai unveiled Tong Tong, which means “little girl” in English, at the Artificial General Intelligence Technology Frontier Exhibition held in Beijing from January 28th to 29th.

    Chinese entity – Bigai – Founded by Song-Chun Zhu in 2020. He lived, worked, and studied in the United States for 28 years before leaving his professorship at UCLA to return home and start a company.

    interpret human intentions

    Ton Ton was programmed to cleanse his environment and keep his surroundings tidy. Ideally, she would be able to “interpret human intentions,” according to the South China Morning Post. (SCMP), due to her ability to know what humans will do next. For example, if Tong Tong finds a bent frame, he straightens it.

    When she sees milk spilled, she cleans it up herself without being asked.

    “Ton-Ton has a heart and strives to understand the common sense that humans teach her. She distinguishes between right and wrong, expresses her attitude in different situations, and has the power to shape the future. ” Bigai said in a video post.

    According to SCMP, Ton Ton depicts the behavior of 3- to 4-year-old children and is expected to improve their skills, values, and knowledge through continuous interaction.

    Also read: AI researchers discover that AI models intentionally refuse instructions

    girls have feelings too

    The company views future developments as follows: A big step towards Achieving artificial general intelligence (AGI), the state where machines can think and reason like humans.

    According to Bigai's video, Tonton is able to express his emotions as he “has his own joy, anger, and sadness.” Researcher and Bigai director Song-Chun Zhu further explained that apart from emotions, having common sense, just like humans, is an important element of the general aspect of intelligence.

    Furthermore, the AI ​​child should be able to “not only complete an infinite number of tasks, but also define new tasks autonomously.”

    “To move towards general artificial intelligence, we need to create entities that understand the real world and have a wide range of skills,” Zhu explained.

    Mr. Zhu is a prominent scholar specializing in the field of AI research, including the field of AI in general. autonomous robotcomputer vision, to name a few. The U.S. Naval Research Laboratory has won some of the highest awards in a variety of fields, including the ONR Young Investigator Award, and works at leading institutions in the field of AI.

    ton test

    Bigai also showcased an AI testing platform known as Tong Test. The test was published by Zhu and his team in the journal Engineering, sponsored by the Chinese Academy of Engineering, last August.

    According to SCMP, traditional AI testing that focuses on task orientation, human identification, and “virtual environments” has limitations. For example, the Turing Test only assesses the level of communication between AI and humans.

    Although virtual environment testing creates a realistic experience, it “tends to oversimplify the physical environment.”

    However, researchers say the Tong Test assesses five aspects: cognition, language, vision, motor and learning.

    “With nearly 100 specialized tasks and more than 50 general tasks, Tong Test provides a complete testing regime for the development of general artificial intelligence,” the institute said in a statement on its website. ” he said.

    “For general AI to be seamlessly integrated, human environmentThey must learn and perform tasks in complex settings based on values ​​and an understanding of cause-and-effect relationships,” Zhu said in a statement.

    For this reason, Zhu said his team proposed Tong Test, a new direction for AI testing that specifically focuses on practical abilities and values.

    “Our research guides the way general AI can learn and improve its capabilities more effectively and safely to better serve human society.”


    Leave a Reply

    Please enter your comment!
    Please enter your name here