The other side of Dalio: Battled testicular cancer, admires Musk, believes adult deaths are three times as costly as infant deaths.

Beating AI News Flash: Anthropic CEO Dario Amodei once had testicular cancer. In 2014, he was diagnosed and underwent chemotherapy, ultimately making a full recovery. This experience was previously little known. On November 5, The New York Times interviewed more than 20 of Dario's friends, classmates, and former colleagues in an attempt to reveal another side of this leading figure in the AI safety camp.

Dario's focus on death and time can be traced back even earlier. In 2006, his father died of hepatitis C, having failed to live long enough for the new drugs that later gave most patients a chance at a cure. Dario originally studied physics, and afterward gradually shifted toward biology, hoping to accelerate medical progress through new technologies. Friends recall that he insisted on rigorous swimming training, tried anti-aging supplements and calorie restriction, and often talked about the finiteness of time, believing that time is the only resource that cannot be regenerated.

As early as 2007, Dario was fascinated by the idea of AI surpassing human intelligence. He once predicted to friends that the technological singularity might arrive in 2040, and that humans might even merge with machines. But he also warned that if AI is not developed correctly, humanity could be doomed.

At the time, he was deeply influenced by AI safety thinker Eliezer Yudkowsky and the effective altruism movement, and he had also corresponded with Musk. Two people familiar with the matter revealed that when Dario was young, he greatly admired Musk and even considered working at his company, though it ultimately did not happen.

In 2010, Dario wrote in a blog post for the charity GiveWell that adults have richer and deeper life experiences than infants, so he believed that the loss caused by the death of one adult could be 2 to 3 times that of the death of an infant.

At the time, he was comparing two charitable projects, one dedicated to preventing and treating adult tuberculosis and the other helping African children get vaccinated. According to his judgment of the value of life, saving adults had the advantage. But in the end, he donated all $10,000 to the child vaccination project. He placed greater value on the latter's reliable implementation process and the long-term impact that rewarding highly effective charities might bring.

To help more people, he even considered giving up scientific research and switching to finance. In 2010, he said in an email that he was considering working in finance because it would allow him to earn more money for donations.

This way of thinking, which believes in rational calculation and values reliable execution, is also reflected in his views on AI. He believes that superintelligent AI could destroy humanity, so it must be developed by people who truly value safety and can be trusted. This strong belief in AI safety gradually evolved within OpenAI into disputes over who has the right to oversee technology and who can participate in core projects.

In 2018, OpenAI researcher Joshua Achiam opposed Dario's appointment as the company's Chief Ethics Officer. He was concerned about the excessive concentration of power in the Amodei siblings and advocated bringing in external experts to prevent a small group from monopolizing ethical oversight of AI. People familiar with the matter recall that Dario raised his voice in person, insisting that no one was more suited to the position than himself.

Around the GPT project, the conflict became even more intense. After Dario and his sister Daniela took the lead in related research, they strenuously blocked OpenAI President Greg Brockman from participating. Dario repeatedly issued ultimatums to Altman: if Brockman was not kept away from the project, he would resign.

Altman ultimately refused to back down. In December 2020, Dario and his sister left OpenAI along with several colleagues, and founded Anthropic four weeks later. The split was also entangled with disputes over research credit, management control, and long-accumulated personal grievances. Dario believed that Altman had downplayed his contributions, and he had also explicitly stated that he could not work with Brockman.

In a striking contrast, after Trump was first elected U.S. president in 2016, Dario expressed concerns in internal OpenAI messages about the direction of U.S. politics. He believed that investor Peter Thiel deserved moral condemnation for supporting Trump, and that OpenAI should not continue accepting his money. But later, Thiel's venture capital firm Founders Fund became an investor in Anthropic.

Today, Anthropic, led by Dario, is pushing for stricter AI safety oversight. He has also publicly clashed with tech giants such as Jensen Huang and Mark Zuckerberg over his advocacy for restricting chip exports to China, strengthening regulation, and guarding against AI going out of control. Supporters believe he is developing AI responsibly, while critics argue that he is trying to give a small number of companies too much technological power.

Dario himself declined to be interviewed for this New York Times article.