Artificial General Intelligence Has Arrived: AI with Human-Level Intelligence
Nvidia CEO and co-founder Jensen Huang has stated that the era of Artificial General Intelligence (AGI) has arrived. This assertion follows OpenAI’s launch of its latest AI model, GPT-6 Astra.
OpenAI introduced GPT-6 Astra last Thursday. The company claims the model brings significant improvements to computer usage, programming, scientific research, and professional tasks.
Astra was initially made available to corporate customers via Daybreak access. OpenAI has since expanded its availability to ChatGPT Plus, Pro, Business, and Enterprise users, as well as through the OpenAI API and Amazon Web Services.
Huang welcomed the launch via a post on the X platform. The Nvidia chief noted that OpenAI’s latest model marks the beginning of the AGI era.
“GPT-6 Astra, trained with approximately 100,000 NVIDIA Grace Blackwell NVLink72. From ChatGPT to o1 to Astra in 4 years. AGI has arrived. Congratulations to the @OpenAI team. The next 400,000 GPUs are coming online soon,” Huang wrote.
The term AGI itself does not yet have a widely agreed-upon definition within the technology industry. Generally, AGI refers to AI systems possessing human-equivalent cognitive abilities in learning and reasoning, capable of acquiring new skills and abilities without requiring retraining.
OpenAI defines AGI as an autonomous system that outperforms humans in most economically valuable tasks.
Meanwhile, Huang and Nvidia utilise a different benchmark. They define AGI as an AI system capable of passing professional certifications and standardised tests across various professional fields with top-tier scores.
OpenAI President Greg Brockman previously described GPT-6 Astra as a pivotal moment in the journey towards AGI. Ahead of the Astra launch, Brockman stated it was reasonable for the public to feel that the AGI era has commenced.
“I do not think it is an exaggeration to feel that we are now in the era of AGI,” Brockman said, as quoted by Fox Business.
OpenAI stated that Astra is capable of performing various computer-based tasks previously considered tedious, such as filling out online forms, updating customer data, managing calendars, and conducting research.
The model can also generate documents, spreadsheets, and presentations. OpenAI even referred to Astra as the “best model for software engineering to date,” while highlighting its cyber security capabilities.
OpenAI CEO Sam Altman also revealed one of Astra’s most impressive capabilities. According to him, the model does not merely execute commands but is also capable of identifying problems that users have not yet considered.
“It is not only capable of conducting extensive research on, for example, the complex chip supply chain we are trying to produce. But as part of that task, which I did not ask for, it can come back and say, ‘You haven’t thought to ask me about another part of this supply chain,’” said Altman.