What is AGI? Jensen Huang's Claims About the New AI Era in the Spotlight

Nvidia CEO Jensen Huang has stated that the era of Artificial General Intelligence (AGI) has arrived. This statement has triggered debate because it is based on a definition of AGI considered looser than the general understanding. AGI is known as a concept of artificial intelligence (AI) with cognitive abilities equivalent to humans, enabling it to learn, understand, and solve various general tasks. Huang’s claim was made during a conversation with US technology podcaster Lex Fridman, where he assessed that current AI, in certain contexts, can already be categorised as AGI. So, what is meant by AGI? Here is a detailed explanation. In essence, AGI is envisioned as AI that can learn, understand, adapt, and solve various problems generally like humans. IBM also emphasises that this concept differs from narrow AI, which is AI designed only for specific tasks such as chatbots, image recognition, or recommendation systems. Thus, AGI is not just clever in one function but has more flexible capabilities across contexts. According to IBM, to date, there is no single definition of AGI that is widely agreed upon. The debate is not only about the technology but also about what is actually meant by “intelligence” in machines. That is why AGI remains a concept that continues to be debated, including when technology industry players use different definitions. This difference in interpretation is evident in the conversation between Lex Fridman and Nvidia CEO Jensen Huang. In that podcast, Fridman described AGI as AI capable of starting, growing, and running a unicorn startup, a technology startup company with a valuation of more than $1 billion USD. This definition is higher and broader because it demands technical skills, business strategy, organisational management, and business sustainability. Meanwhile, Huang used a looser standard. In the 2023 DealBook Summit forum, he once referred to AGI as software capable of competitively passing various human intelligence tests, and at that time estimated it would take about five years to achieve. However, when asked again by Fridman whether AI with that definition could arrive in 5 or 20 years, Huang instead answered that the AGI era has actually already arrived. This claim arises because Huang appears to use a narrower understanding. According to him, AI does not have to be able to build a business that lasts long, does not need to manage people, face boards of directors, or keep a company alive in the long term.