Researchers at the Institute of Automation of the Chinese Academy of Sciences have affirmed that multimodal large language models can self-teach to 'comprehend' concepts during training, mimicking human cognitive processes. This groundbreaking discovery paves the way for investigating artificial intelligence's thinking mechanisms and establishes a foundation for developing future AI systems capable of understanding the world in human-like ways. The pertinent research findings have been published online in the esteemed journal Nature Machine Intelligence.