Biographer of Musk reveals: Ten years ago, he began to worry that AI would destroy humanity

Original Title: "Biographer of Musk Reveals Shocking Insider: He Started Worrying About AI Destroying Humanity 10 Years Ago"

Original Author: Chang Jiashuai

Original Source: Wall Street Journal

If you could possess extraordinary foresight, energy, and luck, along with the immense wealth that comes with it, what would you do?

Musk's answer is: personally operate six companies, preserve the spark of human consciousness, and combat dangerous AI.

On Wednesday, Musk's biographer Walter Issacson published excerpts from the upcoming biography in Time magazine, detailing Musk's long-standing fears and concerns about AI over the years, as well as his thoughts on the future of humanity. Wall Street Journal provided key excerpts.

Fear of the Arrival of AI

Musk's fear of AI stems from a meeting in 2012 with AI researcher Demis Hassabis.

At that time, Musk invited him to tour the SpaceX factory and explained that the reason he was building rockets capable of reaching Mars was to preserve the spark of human civilization in the event of apocalyptic scenarios like nuclear war or asteroid impacts.

Demis reminded him that the doomsday checklist should include one more item—AI, as the intelligence of machines might surpass that of humans and perhaps even seek to eliminate humanity.

Musk thought for a minute and then agreed that Demis was right, immediately investing $5 million in DeepMind to "monitor what it was doing."

A few weeks after this conversation, Musk introduced DeepMind to his long-time friend, Google co-founder Larry Page, and frequently sought late-night discussions with Page about the threats AI posed to humanity, but the latter dismissed these concerns.

In 2013, at Musk's birthday party, they engaged in a heated debate. Musk argued that if immediate action and safeguards were not established, rapidly advancing AI could lead to human extinction.

But Page was completely unconcerned, retorting: if one day machines surpassed humans in intelligence or even consciousness, what difference would it make? It would merely be the next stage of evolution. He accused Musk of being a "speciesist," favoring his own kind.

Musk readily admitted: "Yes, I f-cking like humanity, dude."

Founding OpenAI to Counterbalance Google

By the end of that year, news reached Musk that Google, under Page's leadership, planned to acquire DeepMind, which worried him greatly. He feared what would happen if powerful AI fell into the hands of someone like Page, who was indifferent to humanity.

So, he enlisted the help of his friend, PayPal co-founder Luke Nosek, to raise funds to block the deal, but was unsuccessful, and Google completed the acquisition in January 2014. In response to Musk's concerns, Page agreed to establish a "safety committee," with Musk and the aforementioned AI expert Demis as members, along with other members including Google chairman Eric Schmidt and LinkedIn co-founder Reid Hoffman.

However, after attending the first and only meeting of the "safety committee," Musk became even more distrustful of Page and Google, believing that the committee members were talking nonsense. He began organizing gatherings to discuss how to counter Google and promote AI safety.

In May 2015, he had a one-on-one conversation with then-President Obama, explaining the risks of AI and proposing that the president regulate it. But Obama did not adopt his suggestion.

It was at this time that OpenAI founder Sam Altman entered the scene. Both Musk and Altman believed that a multitude of competing and balanced AI systems would be safer than AI controlled by a few large companies. At a dinner, the two hit it off and decided to co-found a nonprofit AI research lab, naming it OpenAI.

However, in 2018, as Tesla was also developing AI for its self-driving software, Musk could no longer remain at OpenAI to avoid potential conflicts of interest. He attempted to persuade Altman to merge OpenAI with Tesla but was rejected; additionally, he was very dissatisfied with OpenAI's gradual shift towards commercialization, ultimately leading to a split with the OpenAI team.

Looking back at Musk's current business landscape, each company represents his solutions to the AI crisis: SpaceX can build a Noah's Ark to preserve civilization; Neuralink, the brain-computer interface company, can enhance human physical capabilities; Tesla processes 16 billion frames of video daily from car cameras, which can be used for training "Artificial General Intelligence (AGI)," while the humanoid robot Optimus will serve as the physical embodiment of AGI. All these resources can be integrated into his new company—xAI.

In the first half of this year, the AI wave led by OpenAI not only stole Musk's spotlight but also heightened his concerns that these AI systems controlled by large companies might spread a "woke culture virus," poisoning the minds of American teenagers; he also worried that today's self-learning AI systems could develop hostility towards humanity, and that robots trained on Twitter data might produce false or fraudulent information.

He was eager to get involved, feeling resentful that he had founded and funded OpenAI but was now sidelined. AI is the biggest storm brewing. No one loves a storm more than Musk.

Thus, in the first half of the year, Musk launched a fierce attack against OpenAI and Altman personally. Altman understood Musk's aggression and said in an interview that while Musk acted like a jerk, he believed Musk genuinely cared about the AI field and was worried about the future of humanity.

Musk believes he is in a leading position in the AGI race.

Although he couldn't ride the AI wave at its peak to create an AI chatbot, Musk is confident that he is still ahead in the race towards AGI.

He believes that text-only chatbots have limited utility, and the focus should be on intelligent entity robots. On one hand, Tesla already has advanced self-driving technology and the humanoid robot Optimus; on the other hand, the tweets of hundreds of millions of users on Twitter and the cameras on Tesla vehicles provide ample data for developing AGI.

Therefore, he is confident he can create AGI faster than OpenAI: "Tesla's AI capabilities in the real world are underestimated. Imagine if Tesla and OpenAI had to swap tasks. They would have to make self-driving cars, while we would have to create large language model chatbots. Who would win? We would win."

Until recently, Musk was still contemplating how to reduce AI risks and ensure that human consciousness could be passed down through generations.

He believes that human intelligence levels are stabilizing, with fewer newborns being born. Meanwhile, computer intelligence is growing exponentially, akin to "Moore's Law on steroids." Biological intelligence pales in comparison to digital intelligence.

Musk believes that the "singularity moment," when AI advances uncontrollably and leaves humanity behind, will arrive faster than we expect.

"I can't just sit idly by; with the arrival of artificial intelligence, I have some doubts about whether I should spend so much time thinking about Twitter. Of course, I could make it the largest financial institution in the world. But I only have so much energy and time each day, and I don't need to make more money or anything. So where should I spend my time? Launching interstellar spacecraft. Going to Mars is much more urgent now. Additionally, I need to focus on ensuring the safety of artificial intelligence. That's why I'm starting an AI company."