Large language models (LLM) have achieved breakthrough developments in simulating human intuition and natural language expression, yet they are still incapable of focusing on a comprehensive value system and engaging in deep contemplation. Intelligence and consciousness originate concurrently with life, with the evolution of life and the growth of individuals exhibiting continuity. Artifacts, as the materialization of human consciousness, also display continuity, and the development of Artificial General Intelligence (AGI) will unfold along a continuous spectrum. The canxian theory of consciousness and the causal chain re-engineering theory of intelligence provide new directions for intelligence under conditions of smaller model sizes and lower computational power. As AI extends the capabilities of individual humans as our digital avatars, it significantly enhances human capabilities and, in the interaction within the metaverse, promises the formation of superintelligence. Based on this new path to AGI implementation, the existential risks posed by the current mainstream centralized AI large models to humanity can be effectively mitigated.