Artificial General Intelligence (AGI)
Artificial General Intelligence (AGI) is the hypothetical AI system that can perform any intellectual task a human can — reasoning, creating, deciding — without limitation. Unlike today's narrow AI tools, AGI would be self-directing and self-improving. AI companies like OpenAI use AGI as the stated goal of their research, but deliberately refuse to define it precisely, because the honest definition — a machine that thinks like God and can rewrite reality — is, as Professor Jiang argues, the actual secret of the AI project.
Every major AI company — OpenAI, Google DeepMind, Anthropic — lists AGI as their long-term mission. The standard public definition is a machine with human-level general reasoning ability. But when journalist Karen Hao pressed OpenAI founders Greg Brockman and Ilya Sutskever to define exactly what they were building, they could not answer. Professor Jiang's interpretation: they cannot answer because the true definition is that they are trying to create God.
The term was coined by computer scientist John McCarthy, who also invented the phrase "artificial intelligence" specifically to attract funding and public imagination. From the beginning, the terminology was designed to be evocative rather than precise. AGI inherits this ambiguity: it is perpetually being redefined to remain just out of reach, which allows AI companies to perpetually justify new rounds of investment.
From a systems perspective, AGI faces a fundamental constraint. Supervised machine learning — what AI actually is — requires clean data, measurable goals, and defined parameters. AGI, by contrast, would need to handle edge cases with no defined output, make moral judgments with no ground truth, and operate in a world where humans actively resist it. These are not engineering problems to be solved; they are philosophical problems that current AI architecture cannot address.
The political economy of AGI is more revealing than the technology. AI companies claim AGI is imminent to attract trillion-dollar government contracts (Operation Stargate), to justify energy-intensive data center expansion, and to position AI as a national security necessity. The threat that "China will build AGI first" serves the same function as Cold War nuclear anxiety: it generates political will and public funding for a project that cannot justify itself on commercial grounds alone.
Related articles
Related terms
Frequently asked questions
What is Artificial General Intelligence (AGI)? ＋
AGI is the hypothetical AI system capable of performing any intellectual task a human can — without pre-programmed rules for each task. Unlike today's narrow AI (which recognizes faces, translates language, or predicts text), AGI would reason, create, and self-improve across all domains. AI companies like OpenAI claim AGI is their mission, but deliberately refuse to define when AGI has been achieved — giving them permanent justification for expansion and investment.
Why is AGI impossible to achieve according to Professor Jiang? ＋
Because supervised machine learning — the actual technique underlying all AI — requires clean data, measurable goals, and defined parameters. AGI by definition cannot have measurable goals, because its tasks include open-ended moral and philosophical questions. Beyond that, making AI truly universal would require removing all human unpredictability — meaning taking away human autonomy. The real AI apocalypse is not a robot uprising but leaders so convinced AI will save the world that they destroy the world to make AI possible.