# Artificial General Intelligence (AGI) An AI system with human-level cognitive abilities across any intellectual task. Unlike narrow AI, which excels at specific tasks, AGI would generalize across domains without task-specific training. Currently theoretical. Distinguished from [[Strong AI]], which implies consciousness or understanding. AGI focuses on capability breadth rather than inner experience. The term was popularized by Shane Legg and Ben Goertzel. Current [[Large Language Models (LLMs)]] demonstrate impressive generality but lack true cross-domain reasoning, embodied understanding, and autonomous goal-setting that AGI would require. The path to AGI remains deeply contested, with disagreement on timelines, architectures, and whether current approaches can ever get there. AGI development raises fundamental questions around [[AI Safety]] and [[AI Alignment]]. An AGI system that optimizes for the wrong objective at human-level capability represents an existential-class risk. [[AI Scaling Laws]] suggest raw compute alone may not be sufficient. ## References - Term popularized by Shane Legg (DeepMind co-founder) and Ben Goertzel ## Related - [[Strong AI]] - [[Large Language Models (LLMs)]] - [[AI Safety]] - [[AI Alignment]] - [[AI Scaling Laws]]