Interview with Elon Musk: AI, Space Exploration, and the Future of Humanity

Recently, the American startup accelerator Y Combinator held its first AI startup school in San Francisco, inviting several heavyweight figures from the AI industry, including Elon Musk.

In an interview, Musk talked about his recent experience working in government. He compared this experience to "cleaning the beach," while the upcoming AI is like a "thousand-foot-high tsunami." He believes that the work of government efficiency departments is not very significant compared to the enormous changes brought by AI.

Musk predicts that digital superintelligence may arrive this year or next, surpassing human intelligence. He also forecasts that the number of humanoid robots in the future will far exceed that of humans, possibly reaching 5-10 times the human population. He even boldly predicts that the scale of the AI-driven economy will be thousands or even millions of times larger than it is now, with human intelligence potentially dropping below 1%.

Here are some key points from Musk's speech:

• Predictions suggest that digital superintelligence may arrive this year or next, and it will be smarter than humans.

• The number of humanoid robots in the future will far exceed that of humans, potentially being 5 times or even 10 times that of humans.

• The predicted scale of the AI-driven economy will be thousands or even millions of times larger than the current one, pushing civilization towards a higher level, with human intelligence potentially dropping to below 1%.

Elon Musk emphasizes that "a rigorous adherence to the truth" is the most important cornerstone of AI safety, and forcing AI to believe in untruths is extremely dangerous.

Musk also talked about his company xAI training the Grok 3.5 model, focusing on reasoning capabilities. He revealed that xAI has built a large-scale GPU training cluster, including 150,000 H100s, 50,000 H200s, and 30,000 GB200s, with a second data center set to go live with 110,000 GB200s.

Regarding the future of artificial intelligence, Musk predicts that there will be several deep intelligences, possibly between 5 to 10, with about 4 of them in the United States. He believes that these AIs may discover new physics, invent new technologies, but they may also attack each other.

Musk also discussed the importance of becoming a multi-planet species. He predicts that within the next 30 years, humanity may transfer enough material to Mars to make it self-sufficient. He believes this will greatly extend the expected lifespan of civilization or consciousness.

Finally, Musk emphasized that the most important thing for AI safety is to strictly adhere to the truth, even if that truth is politically incorrect. He encouraged young tech talents to do useful things, be as helpful as possible to their peers, and focus on creating super-realistic AI.