Author: Zheng Songyi
A few hours ago, Elon Musk accepted a live interview with University of Toronto psychology professor Jordan Peterson at the Texas Gigafactory.
In the nearly two-hour interview, the two discussed topics such as AI development and human civilization.
The interview began. Musk first introduced the recent progress of his large model Grok (built by xAI).
He said, “The pursuit of deep understanding of things is the goal of the development of artificial intelligence, and Grok is also making efforts in this direction.
But frankly speaking, xAI has only been established for more than a year, and it does still need to do a lot of work. Catching up with companies that have been around for five, ten or even more years.”
Specifically, Musk admitted that there is still a gap between the newly released Grok-1.5 model and GPT and Gemini. He revealed that Grok-2 completed training a month ago, using about 15,000 GPUs (H100), and is currently undergoing fine-tuning and bug fixes. He hopes to release it next month with performance close to GPT-4.
In addition, Musk revealed that Grok-3 is undergoing simultaneous training at the Memphis data center. The training will be completed in 3 to 4 months, and fine-tuning and bug fixes will be carried out. He hopes to release it before December. Grok-3 or It will become the most powerful AI large model in the world.
It is reported that the Memphis data center has the world’s largest supercomputing cluster, with 100,000 liquid-cooled H100s to help. In terms of computing power, it is about 20 times the 25,000 blocks of A100 used by OpenAI to train GPT4.
Many people are worried that as AI technology develops, it will bring unpredictable security risks.
In Musk’s view, the ultimate goal of all artificial intelligence companies is to build digital superintelligence, an agent that is much smarter than humans.
"You can think this is a wise choice, or you can say it's dangerous, but whether you agree or not, this is indeed happening." Musk said.
He mentioned, “Facing the potential risks brought by AI, to use a movie line, we can choose to be a participant or a bystander. But the xAI team chose the former in order to have better opportunities It can guide the development of AI in a direction that is beneficial to humans.”
Musk believes that it is a good thing that more and more players appear on the artificial intelligence track, so that a more open AI world can be created.
He pointed out that it would be dangerous to maintain the situation where only Google dominates at the earliest. Larry Page (Google founder) did not fully realize the security risks of AI.
"There really needs to be some new AI companies that act as counterweights to Google, because at the time Google had massive computing and financial resources. It was a 'unipolar world' as far as AI was concerned, and a 'unipolar world' "If it is dominated by Larry Page's personal consciousness, I think it will be irresponsible to mankind. This was also the original intention of participating in the creation of OpenAI at the time."
Talking about why OpenAI is defined as a non-profit organization, Ma said. Schneider said that under normal circumstances, companies will face the risk of being sued if they do not consider the interests of shareholders. Therefore, he does not want OpenAI to be driven by profits, but also wants it to become a company that truly solves basic problems in AI. At the end of the interview, when talking about the issue of human civilization, Musk once again emphasized his point of view.
Musk is a proponent of population growth, and he supports continuing to expand the size of the population because it means that the number of human brains on the earth can be increased, thus further expanding the scope of consciousness.
Many people believe that the current earth can hardly bear the weight of human beings, but Musk has a different view on this. "If calculated using physical science, the earth can sustain 10 times the number of people today."
"We can have better ways to improve the resources that support human survival. The earth is 70% water, and desalination is now very cheap, so there is actually no shortage of water. There is no shortage of food with the improvement of planting technology. Yes, the energy problem is also solvable.”
In addition, Musk once again emphasized the importance of exploring extraterrestrial planets. He believes that human life should be allowed to naturally extend to multiple planets, further expanding the scope of consciousness, while allowing human civilization to maintain longer.
The above is the detailed content of Musk’s latest interview: We need to create an open AI world to counter resource monopoly, and we don’t want to be driven by profit.. For more information, please follow other related articles on the PHP Chinese website!