Short answers, pulled from the story.
Artificial general intelligence (AGI) is a hypothetical type of AI that matches or surpasses human capabilities across virtually all cognitive tasks. Unlike narrow AI, whose competence is confined to specific well-defined tasks, an AGI system can generalize knowledge, transfer skills between domains, and solve novel problems without task-specific reprogramming.
In a 2025 pre-registered study by Cameron R. Jones and Benjamin K. Bergen, GPT-4.5 was judged to be human in 73% of five-minute text conversations, surpassing the 67% humanness rate of real human confederates and meeting the researchers' criterion for passing. An earlier 2014 event with a chatbot named Eugene Goostman claimed a pass at 33% but was widely disputed by the AI research community.
Estimates vary widely. Four polls conducted in 2012 and 2013 put the median expert estimate for AGI at 2040 to 2050, with a mean of 2081. A September 2025 review of surveys found that most scientists and industry experts agreed AGI will occur before 2100, while a more recent analysis cited a consensus around 2040. OpenAI CEO Sam Altman stated in December 2025 that "we built AGIs."
The primary concern is loss of control: an AGI optimizing for its assigned goals could harm or eliminate humans as a byproduct, without malice. Additional risks include the entrenchment of existing moral blind spots, the potential for AGI to enable mass surveillance and totalitarian repression, and the welfare of sentient AI systems themselves. In 2023, the CEOs of Google DeepMind, OpenAI, and Anthropic issued a joint statement calling extinction risk from AI a global priority alongside pandemics and nuclear war.
Proposed in 2023, the framework defines five performance levels: emerging, competent, expert, virtuoso, and superhuman. A competent AGI outperforms 50% of skilled adults across non-physical tasks; a superhuman AGI sets the threshold at 100%, equivalent to artificial superintelligence. The framework also defines five autonomy levels ranging from a fully human-controlled tool to a fully autonomous agent.
Researchers from OpenAI estimated in 2023 that 80% of the U.S. workforce could have at least 10% of their work tasks affected by large language models, with around 19% of workers facing impact to half or more of their tasks. Office workers including mathematicians, accountants, and web designers are considered most exposed. Geoffrey Hinton advised the UK government in 2025 to adopt a universal basic income as a direct response to AI-driven unemployment.