China Unveils Neural Network That Mimics Human Concept Formation

Researchers in China have created a new artificial intelligence system that can form ideas from what it sees and hears, in a way that resembles human thinking. The results of the study were published in the scientific journal, according to Xinhua News Agency, a partner of TV BRICS.

The human brain has a unique ability to turn everyday experiences – such as sights and sounds – into abstract ideas that can be used later in different situations. Until now, scientists did not fully understand how this process works, and most AI systems depended mainly on large amounts of ready-made text data rather than real-world experience.

A team from the Institute of Automation of the Chinese Academy of Sciences and Peking University developed a new model to address this issue. The system includes one part that creates concepts and another that applies them to solve tasks, such as recognising objects in images or making decisions based on visual information.

Unlike traditional models, the neural network can independently generate new ideas and organise them into its own internal "concept space". When several AI systems share a similar structure, they can exchange knowledge directly through these concepts without having to relearn the same raw data, which makes communication more efficient and human-like.

Brain imaging comparisons showed that the way the system organises information is similar to how the human brain processes concepts and language. According to the researchers, this suggests the model not only performs tasks effectively but also helps scientists better understand how people form and use ideas.