Artificial general intelligence (AGI) refers to a hypothetical type of AI that could match or surpass human capabilities across virtually all cognitive tasks, unlike narrow AI, and it is a major goal for tech giants like OpenAI, Google, and Meta. The term was coined in 1997 by Mark Gubrud and popularized around 2002 by Shane Legg and Ben Goertzel, with a 2020 survey identifying 72 active AGI projects across 37 countries. In 2023, Google DeepMind proposed a classification framework defining five performance levels—from emerging to superhuman—and five autonomy levels, ranging from fully human-controlled tools to fully autonomous agents. Such systems are expected to master reasoning, planning, learning, and natural language, but experts remain divided on whether AGI poses an existential threat, with some urging global priority on risk mitigation while others see it as too remote. Related ideas include artificial superintelligence (ASI)—intelligence far exceeding humans—and transformative AI, which could alter society on the scale of an industrial revolution.