The development of artificial general intelligence (AGI), which refers to AI systems with capabilities on par with humans in most tasks, is a topic that continues to spark debate among scientists. From differing opinions on whether AGI is far in the future or just around the corner, to the belief that today’s large language models (LLMs) are already a form of AGI, there are many perspectives to consider.
In an effort to bring clarity to this discussion, a team of scientists at Google DeepMind, including Chief AGI Scientist Shane Legg, have proposed a new framework for classifying the capabilities and behavior of AGI systems and their precursors. In their paper, they emphasize the need for the AI research community to reflect on what is truly meant by “AGI” and to quantify attributes such as performance, generality, and autonomy of AI systems.
