Recent research from Chinese scientists shows that large language models (LLMs) are beginning to build systems for understanding and classifying natural objects that closely resemble human thought. It’s a fascinating development that nudges us to rethink where machine learning and human cognition meet.
Published in Nature Machine Intelligence, the study explores how we naturally conceptualise and categorise the world around us. By tapping into vast collections of text—and sometimes visual and auditory data—these models are learning to mirror the way we perceive our environment. This raises the intriguing possibility that, one day, AI could tackle complex tasks with a human touch.
The implications are significant. If AI can adopt methods similar to our own way of thinking, we might soon see tools that better understand our needs and work alongside us in more intuitive ways.
