Unraveling the AGI Mystery: Seeking Consensus Amidst OpenAI Drama

Researchers at DeepMind are seeking consensus on the definition of Artificial General Intelligence (AGI). With various interpretations of AGI, the team aims to establish a common language for researchers by outlining a framework for classifying AGI capabilities and behavior. They emphasize that achieving AGI does not imply qualities such as consciousness or sentience. The team also introduces a list of intelligence thresholds ranging from "Level 0, No AGI" to "Level 5, Superhuman," with only three generative text models meeting the criteria for AGI at the "Level 1, Emerging" stage. The researchers acknowledge the need for a living benchmark that includes a framework for generating and agreeing upon new tasks as our understanding of AGI progresses.