Functional AGI refers to a potential class of AI that is fully autonomous and capable of outperforming humans in conducting economically valuable work which involves complex problem-solving, logical reasoning, learning from experience, and adaptation to new situations. Unlike narrow AI, which excels at specific tasks, Functional AGI could generalize knowledge from one domain to another, similarly to how humans can apply learning from past experiences to new problems, tasks, or domains.
The concept of AGI dates back to the beginning of the AI field itself in the 1950s and 60s. However, the term "Artificial General Intelligence" was coined in 2005 by AI researcher Ben Goertzel, and has grown in popularity since amid debates about the future direction of the AI field.
Notable contributors to the development and theoretical exploration of AGI include Ben Goertzel, who coined the term and established the AGI research field, along with pioneers like Marvin Minsky and John McCarthy, who envisioned machines capable of general intelligent action early in the history of AI.