Jensen Huang (NVIDIA) claims AGI has been achieved

In a discussion at the Stanford Institute for Economic Policy Research, NVIDIA CEO Jensen Huang made a provocative claim about the state of Artificial General Intelligence (AGI). He suggested that AGI could be considered "achieved" in as little as five years, depending on how the goal is defined. Huang's argument centers on a pragmatic, test-based definition: if an AI system can pass a set of comprehensive human exams—such as legal bar tests, logical reasoning assessments, or specialized academic qualifiers—it should be recognized as possessing human-level intelligence for that domain. This shifts the AGI benchmark away from philosophical debates about consciousness and toward measurable, expert-level performance on specific tasks.

Huang's statement is significant because it comes from the leader of the company whose GPUs power the vast majority of advanced AI training. He framed the AGI question as an engineering challenge of expanding an AI's skill set and knowledge base. By focusing on exam performance, Huang is effectively redefining the finish line for one of tech's grandest goals. This perspective suggests that milestones like OpenAI's GPT-4 passing the bar exam or Google's models excelling at medical licensing tests are not just incremental improvements but steps toward a redefined AGI. His comments have ignited debate about whether AGI should be measured by breadth of general reasoning or by depth of expert capability.