According to Stephen Smith, who serves as both the President of the Association for the Advancement of Artificial Intelligence and a Professor at the Robotics Institute of Carnegie Mellon University in the United States, the future trajectory of artificial intelligence (AI) development might well mirror the human learning process—embracing continuous, or even 'lifelong,' learning. Smith highlighted that current large language models come with inherent limitations. They lack the capability to consistently update and 'evolve' over time, as well as the proficiency for causal reasoning. He posited that AI could adopt 'learning methods' akin to those of humans, leveraging small-sample, meticulously curated datasets, and engaging in active experimentation to attain 'lifelong learning.' Moreover, strides in embodied intelligence, exemplified by robots, have the potential to propel AI development forward by accumulating experiences through interactions with the physical world and bridging 'knowledge gaps.'