Nvidia CEO Jensen Huang said OpenAI's new Astra model marks the arrival of artificial general intelligence, pointing to the model's training on Nvidia's chips. AI researcher Gary Marcus and the ARC Prize benchmark team pushed back on the claim, arguing Astra falls short of any agreed AGI definition.
Nvidia CEO Jensen Huang said Sunday that OpenAI's newest model has crossed into artificial general intelligence, a claim that drew pushback from AI researchers. Huang wrote on X that "AGI has arrived. Congratulations @OpenAI team." He also pointed out that the model, called Astra, was trained on Nvidia's chips.
OpenAI's Astra debut
OpenAI unveiled Astra on Thursday, calling it the world's "most intelligent and aligned model" and saying it can perform demanding professional work with unmatched speed and accuracy. OpenAI president Greg Brockman told reporters on the same call that the company had entered the AGI era and said he personally believes it has reached that threshold. OpenAI defines AGI as highly autonomous systems that outperform humans at most economically valuable work, and it began rolling Astra out to customers this week.
AGI claim draws pushback
Huang made a similar claim in March, when podcaster Lex Fridman asked him about an AGI benchmark tied to running a company worth more than $1 billion, and Huang said he believed it had already been achieved. AI researcher Gary Marcus wrote on his Substack Sunday evening that Huang offered no evidence and no definitions, calling it an attempt to settle a scientific question by corporate fiat, and said Astra meets only one or two points of his own ten-point AGI definition.
The team behind the ARC Prize benchmark, which OpenAI cited when announcing Astra, also stopped short of calling the model AGI, saying its tests run in bounded environments that do not reflect the complexity of the real world. OpenAI CEO Sam Altman has separately called AGI a poorly defined term.
Nvidia's chip windfall
Frontier AI developers including OpenAI, Meta, Anthropic, and Google have grown increasingly reliant on Nvidia's chips, and OpenAI has called Nvidia the foundation of its infrastructure. That demand showed up in Nvidia's results: the company reported $96.2 billion in quarterly revenue in August, more than double the year-earlier period, with its data center business generating $89 billion of that total.
In the same Sunday post, Huang said more capacity is coming, pointing to 400K GPUs set to come online next.
Source: Business Insider
Trading involves risk.
Risk Disclaimer
