Huang Declares AGI Has Arrived in Practical Tasks

Nvidia CEO Jensen Huang has made headlines by declaring that artificial general intelligence (AGI) is already here, not just in theory but in practice. Speaking on the Lex Fridman podcast, he stated 'I think it's now. I think we've achieved AGI.'

This stance may come as a surprise to some, especially given the long-standing consensus among AI researchers that achieving true AGI is still a distant goal. However, Huang's definition of AGI is different from what many experts would consider the traditional benchmark for measuring an AI system's capabilities.

Huang argues that instead of focusing on whether an AI can surpass human cognition across the board, we should be looking at its ability to generate 'profitable tokens,' or productive economic output. This means that if an AI system is capable of producing something valuable and contributing to a company's bottom line, then it has achieved AGI in his eyes.

Nvidia is already working towards a model where over 400,000 AI agents operate alongside roughly 40,000 human employees, handling tasks such as engineering, operations, and business functions. The company envisions these agents as an integral part of its workforce, with the goal of producing intelligence as an industrial output.