The battle for AI in Silicon Valley: The pinnacle of human innovation driven by ambition, fear, and money

The New York Times article on December 7, "Ambition, Fear, and Money: How the Battle for AI in Silicon Valley Was Ignited," unveiled a heated debate between Musk and Page eight years ago that sparked a fierce competition for control of AI and gave birth to OpenAI. But now, the people who are most worried about the risks of AI are deciding that it is up to them to shape it. As a result, mistrust fuels competition.

SAN FRANCISCO — Elon Musk celebrated his 44th birthday in July 2015 when his wife threw a three-day party for him at a winery resort in California. Invited are family and friends, and this upscale hotel in Napa Valley consists of many cabin rooms where children run and play.

It was years before Twitter became X and Tesla became profitable, and a year before Musk and his wife, Tarulla Riley, ended their second marriage. Party guest Larry Page was still Google's CEO at the time. And just a few years ago, artificial intelligence was just coming into the public eye, and it was used to identify cats on YouTube with an accuracy rate of 16%.

After dinner on the first day, Musk and Page sat down by the fire by the pool to delve into the topic of artificial intelligence. The two billionaires have been friends for more than a decade.

But on that clear night, the conversation quickly became heated, and the two disagreed on the question of whether AI would ultimately promote human progress or destroy it.

Comics: Musk vs. Larry Page

As the chill of the night set in, the atmosphere of discussion became more and more lively, and more than 30 partygoers gathered to listen to their conversation. For more than a decade, Paige's vocal cords have been plagued by an unusual illness, and he whispered about his vision of a digital utopia. Humans will eventually merge with AI machines, he said. One day, there will be a variety of intelligent competition for resources, and the strong will survive.

If that happens, Musk says, we're done. Machines will destroy humanity.

With a hint of frustration, Page insisted that his utopia was worth fighting for. Finally, he called Musk a "speciesist," the kind of person who favors humans over future digital life forms.

Musk later said that the slander became "the last straw."

Musk and Larry Page back then

Eight years later, the argument between the two seems quite predictable. The question of whether AI will enhance or destroy the world — or at least wreak havoc — has sparked an ongoing debate among Silicon Valley founders, chatbot users, academics, lawmakers and regulators about whether AI technology should be regulated or left to its own devices.

The debate pits some of the world's richest people against each other: Elon Musk, Page, Mark Zuckerberg of the "metaverse" platform company, Peter Thiel, tech investor Peter Thiel, Microsoft's Satya Nadella, and OpenAI's Sam Altman. All are fighting for a piece of the pie and the power to shape it.

At the heart of this controversy is a puzzling paradox. Those who say they are most worried about AI are also the ones who are most active in promoting its development and profiting from it. They believe that their grand vision is justified on the basis of a firm belief that only they can stop AI from harming the planet.

Shortly after the party that summer, Musk and Page stopped dating. A few weeks later, in a private room at the Rosewood Menlo Park Hotel in California, Mr. Musk had dinner with Mr. Altman, who was running a tech incubator at the time, and several researchers.

That dinner led to the launch of a startup called OpenAI later that year. With hundreds of millions of dollars in support from Musk and other investors, the lab has pledged to protect the world from Page's vision.

With the ChatGPT chatbot, OpenAI has fundamentally changed the tech industry and shown the world the risks and potential of AI. According to two people familiar with OpenAI's latest funding round, the company is valued at more than $80 billion, but the partnership between Musk and Altman has not lasted. The two have not been in contact with each other since.

"There are reasons for disagreement, mistrust and egotism," Altman said. "The more people move in the same direction, the more divergent they become. The same is true in religious groups and sects. The closest people tend to quarrel more fiercely with each other. ”

Last month, that infighting came to OpenAI's board of directors. The rebellious board members tried to drive Altman away, believing they could no longer trust him to build artificial intelligence for the benefit of humanity. During five chaotic days, OpenAI looked untenable, until the board finally backed down under pressure from big investors and employees who threatened to follow Altman in his departure.

The farce within OpenAI has given the world its first glimpse of a fierce battle between those who will determine the future of artificial intelligence.

But years before OpenAI's near-collapse, there was a little-known fierce competition in Silicon Valley for control of the technology that is rapidly reshaping the world. The New York Times interviewed more than 80 executives, scientists and entrepreneurs, including two who attended Musk's birthday party in 2015, to tell this story of ambition, fear and money.

Birth of DeepMind

Five years before the Napa Valley party, and two years before the cat breakthrough on YouTube, 34-year-old neuroscientist Demis Hassabis walked into a cocktail party at Thiel's San Francisco mansion and immediately knew there was an opportunity. There is a chess board in Thiel's living room. Hassabis was once the second player in the under-14 category in the world chess game.

"I prepared for that meeting for a year," Mr. Hassabis said. "I thought it would be a stepping stone for me: I knew he loved to play chess. ”

In 2010, Hassabis and two colleagues living in the United Kingdom were looking for funding to create artificial general intelligence, a machine that could do anything the brain could do. At the time, few people were interested in artificial intelligence.

However, some scientists and thinkers have begun to pay attention to the risks of AI. Many, including the three young men from the UK, are in contact with Eliezer Yudkowski, an internet philosopher and self-taught artificial intelligence researcher. Yudkowski was the leader of a group of self-proclaimed rationalists who later called themselves effective altruists.

Because of his early investment in Facebook and his early partnership with Musk at PayPal, Thiel became very wealthy. He developed a keen interest in the "singularity," a science fiction metaphor used to describe the moment when intelligent technology is no longer under human control.

Yudkowski used funding from Tyll to expand his AI lab and start an annual conference on the theme of the "singularity." Yudkoski had met with Yudkoski many years ago by one of Hassabis's colleagues, and Yudkoski arranged for the three of them to speak at the meeting, ensuring that they would be invited to the party in Till.

Yudkovsky introduced Hassabis to Till.

Teal was captivated and invited the three of them to the house again the next day. The trio presented their plans, and soon Teal and his venture capital firm agreed to invest £1.4 million ($2.25 million) in their startup. Thiel was their first major investor.

They named the company DeepMind, a reference to the so-called "deep learning" of artificial intelligence systems that Xi skills by analyzing large amounts Xi of data, neuroscience, and the "deep thinking" supercomputer in the science fiction novel The Hitchhiker's Guide to the Galaxy. By the fall of 2010, they began to build their dream machine. They genuinely believe that because they understand the risks, they are the only ones who can protect the world.

"I don't think it's a contradictory position," said Mustafa Suleiman, one of the three founders of DeepMind. "These technologies will bring tremendous benefits. The goal is not to exclude them or suspend their development, but to reduce the disadvantages. ”

After gaining the support of Till, Hassabis began to work hard to get into Musk's orbit. About two years later, the two met at a meeting organized by Thiel's investment fund, which had also invested in Musk's SpaceX. Hassabis was given the opportunity to visit SpaceX's headquarters. After that, the two had lunch together in the company cafeteria and had a conversation.

Musk explained that his plan is to colonize Mars in order to escape the problem of overpopulation and other dangers on Earth. Hassabis replied that the plan would work, as long as super-intelligent machines did not follow Mars to destroy humanity.

Musk plans to colonize Mars

Musk was speechless. He had not yet considered this unusual danger. Soon, Musk joined Thiel and began to invest in DeepMind.

DeepMind, which is well-funded, has hired people who specialize in neural networks. A neural network is a complex algorithm that mimics the processes of the human brain and is essentially a huge mathematical system that recognizes patterns in large amounts of numerical data, and it sometimes takes days, weeks, or even months to recognize a pattern. The earliest neural networks were developed in the 50s of the 20th century and can be used to learn Xi handle tasks on their own. For example, after analyzing scribbled names and addresses written on hundreds of envelopes, the neural network was able to recognize handwritten documents.

DeepMind takes this concept a step further. In order to show what neural networks can do, the company built a system that can learn Xi play the classic game "Atari".

This caught the attention of Google, another giant in Silicon Valley, especially Page.

Talent Auction

In the fall of 2012, Jeffrey Hinton, a 64-year-old professor at the University of Toronto, Canada, and two graduate students published a research paper demonstrating the capabilities of artificial intelligence. They trained a neural network to recognize common objects, such as flowers, dogs, and cars.

Scientists were amazed by the accuracy of the technique developed by Hinton and his students. Their work has particularly caught the attention of artificial intelligence researcher Yu Kai. Mr. Yu, who had recently started working for the Chinese internet giant Baidu, had met Mr. Hinton at an academic conference. According to three people familiar with the matter, Baidu offered to offer Hinton and his students $12 million to join the Beijing-based company.

Hinton declined Baidu's offer, but the money caught his attention.

"We didn't know how much we were worth," Hinton said. He consulted with lawyers and experts involved in the company's acquisition and came up with a plan: "We are going to organize an auction and sell ourselves." "The auction took place during the annual meeting of artificial intelligence held at the Lake Tahoe Resort & Casino Hotel in Heras.

Google made the bid. Microsoft also made a bid. DeepMind quickly pulled out. The industry giants raised their bids to $20 million, then to $25 million. When the bid exceeded $30 million, Microsoft pulled out, but re-entered the auction when the bid reached $37 million.

Later, Microsoft withdrew for the second time. Only Baidu and Google remained, pushing the bid to $42 million, $43 million. When the bid reached $44 million, Hinton and his students finally stopped the auction. While the bidding is still rising, they want to work for Google. Moreover, the price is already incredible.

It's a surefire sign that deep-pocketed companies are determined to buy up the most talented AI researchers. This did not escape the attention of DeepMind's Hassabis. He has always told employees that DeepMind will remain an independent company. He believes that is the best way to ensure that the company's technology does not become dangerous.

But as Big Tech jumped into the talent race, Hassabis made a decision that he had no choice: the time had come to sell the lab.

By the end of 2012, both Google and Facebook were considering buying the London-based lab, according to three people familiar with the matter. Hassabis and the lab's co-founders insist on two conditions: that no DeepMind technology be used for military purposes, and that its general AI technology must be overseen by an independent technical and ethics committee.

Google made a bid of $650 million. Facebook's Mark Zuckerberg offered DeepMind's founders a higher price, but he disagreed to either condition. DeepMind chose Google.

The Disappearing Ethics Committee

When Musk invested in DeepMind, he broke his unwritten rule that he would not invest in any company that he did not run himself. Less than a month after his spat with Page at his birthday party, the downside of his decision to invest in DeepMind became apparent when he sat down face-to-face with his former friend who is also a billionaire again.

That was when the DeepMind Ethics Committee held its first meeting on August 14, 2015. The committee was formed at the insistence of the startup's founders to ensure that their technology doesn't cause any harm once it's sold. According to three people with knowledge of the meeting, members of the committee convened the meeting in a conference room outside Musk's SpaceX office.

But that's where Musk's control ends. Google bought DeepMind when it bought the entire company. Musk is out.

Three Google executives who now firmly control DeepMind were all present: Page, Google co-founder and Tesla investor Sergey Brin, and Google Chairman Eric Schmidt. Other attendees included Reed Hoffman, one of PayPal's founders, and Toby Odeh, an Australian philosopher who studies "risks associated with human existence."

At the meeting, the founders of DeepMind reported on the work they were moving on, but they realized that the technology posed serious risks.

Eight months later, DeepMind has made significant progress that has shocked the AI community and the world. A DeepMind machine called AlphaGo beats one of the best Go players in the world. 200 million people around the world watched the match, which was broadcast live on the Internet. Most researchers previously thought it would take a decade for AI to get Go to this level.

Larry Page

Part company with

Convinced that Page's optimistic view of AI was completely wrong and angry at the loss of DeepMind, Musk decided to set up his own lab.

At the end of 2015, OpenAI was founded.

In late 2017, Musk hatched a plan to take control of OpenAI from Altman and other founders, according to four people familiar with the matter, and he wanted to turn the lab into a commercial entity that could join forces with Tesla and run on a supercomputer being developed by the car company.

After being opposed by Altman and others, Musk said he would quit the lab and said he would focus on Tesla's own AI system. In February 2018, Musk announced to OpenAI employees that he was leaving from the top floor of the startup's office building converted from a truck factory, three people who attended the meeting said.

OpenAI is suddenly in dire need of new funding. Altman met Microsoft CEO Satya Nadella while flying to Sun Valley for a conference. Cooperation seems to be a natural thing to do. The two sides reached an agreement in 2019.

Altman and OpenAI, which formed a for-profit company under the original nonprofit framework, received $1 billion in new capital, and Microsoft had a new way to incorporate AI into its vast cloud computing services.

Reveal

After OpenAI received another $2 billion in funding from Microsoft, Altman and another executive, Greg Brockman, visited Bill Gates at his mansion on Lake Washington outside Seattle.

Gates told them over dinner that he was skeptical about the effectiveness of large language models. He said he would remain skeptical until the technology could complete tasks that required critical thinking, such as getting a passing grade on a biology pre-university exam.

Five months later, on August 24, 2022, Altman and Brockman came to Gates' house again, bringing with them an OpenAI researcher named Chelsea Wirth. Voss won a medal in the International Biology Olympiad in high school.

On a giant digital display outside Gates' living room, OpenAI staff demonstrated to him a technology called GPT-4.

Brockman submits an advanced biology multiple-choice test paper to the system, and Voss scores the machine's answers.

There are a total of 60 questions. GPT-4 only got one wrong answer.

Gates, who was sitting in his chair, straightened up and his eyes widened. He had a similar reaction when researchers showed him a graphical user interface in 1980. The graphical user interface is the foundation of the modern personal computer, and Gates believes GPT is that revolutionary.

By October 2022, Microsoft had added GPT technology to its online services, including the Bing search engine. Two months later, OpenAI released the ChatGPT chatbot, which is now used by 100 million people on a weekly basis.

OpenAI beats out the effective altruists of Anthropic. While Google's Page optimists rushed to release their own chatbot, Bard, it was widely believed that they had lost the race against OpenAI. Three months after ChatGPT's release, Google's stock price fell 11%. There is no trace of what Musk has done.

But that's just the beginning.

Bill gates