Before Leaving OpenAI, Anthropic President Had a 'Stuffed Animal Advisory Board'

Author: Li Jia, Wall Street Journal

Today, Anthropic has made "AI safety" its core label, but few know that the company's co-founder and president, Daniela Amodei, once discussed management issues with her husband, Holden Karnofsky, using a group of stuffed animals.

These toys were given different personalities: some were responsible for "having fewer meetings and avoiding unnecessary trouble," while others emphasized rules, details, and execution, and some were tasked with reminding decision-makers to maintain empathy. When faced with a tricky workplace problem, Daniela and her husband would let these virtual characters "voice their opinions," simulating the choices that different types of managers might make.

This rather personal, even somewhat whimsical anecdote occurred during Daniela's time at OpenAI. During this period, she and her brother, Dario Amodei, were gradually developing differences with OpenAI's management: Daniela experienced friction with several senior executives, while Dario became increasingly wary of the company's leadership and the direction of advanced AI development.

In 2020, the siblings ultimately decided to leave OpenAI, a split later referred to internally as a "divorce," which also became the starting point for the birth of Anthropic.

The Atlantic recently cited journalist Kevin Roose's upcoming book, "The AGI Chronicles," which reveals this little-known experience. Daniela stated in a 2023 interview with Roose that this method "is very useful for analyzing different personality types."

From forming a "consulting committee" with stuffed animals to growing into an AI giant with thousands of employees, Anthropic has evolved from a small AI safety startup to an important player in the industry. According to media reports, the company's potential IPO valuation could approach $2 trillion.

How a Group of Stuffed Animals Became a "Consulting Committee"?

This informal "advisory group" was not just a simple joke, but a way to simulate different decision-making styles.

Among them, Jinji is an easygoing cat that tends to lighten Amodei's workload, such as skipping unnecessary meetings; the pink bear Maura represents a completely opposite style—organized, demanding, and detail-oriented; the panda named Beary Bonds is set to add more empathy and generosity to discussions.

When faced with the same workplace issue, the Amodeis would let the "personalities" represented by different toys respond accordingly.

The real function of this method lies not in the toys, but in forcing decision-makers to temporarily step away from their own positions and re-examine problems from another personality's perspective.

Essentially, this is a low-cost thought experiment: without real external consultants, one can break down the same problem into different decision paths through simulations of different roles.

And this method was almost unknown within OpenAI.

Beyond Stuffed Animals, Divisions Within OpenAI Are Widening

Meanwhile, the relationship between the Amodei siblings and OpenAI was also changing.

Daniela Amodei joined OpenAI in 2018, later transitioning to a management role in the technical team, where she had disagreements with several senior executives, including Greg Brockman and Mira Murati.

She was once close to Brockman, who gave a speech at her wedding to Holden Karnofsky in 2017, but their working relationship gradually deteriorated thereafter.

Dario Amodei became increasingly concerned about OpenAI's leadership and the direction of advanced AI, growing wary of Sam Altman and Brockman.

At the same time, he began to privately organize activities with a group of researchers who wanted to place greater emphasis on the risks of powerful AI systems. Outside the normal company structure, these individuals communicated and held meetings, with some colleagues referring to this faction as "Dario's little circle."

The internal conflict ultimately reached an unavoidable stage.

In 2020, Dario began to assemble a team willing to leave OpenAI to advance advanced AI research with different safety philosophies. Daniela became one of the core figures in this exodus, responsible for negotiating with OpenAI during the team's preparations to leave.

This split was later referred to privately by insiders as a "divorce."

An Exodus That Gave Birth to an AI Giant

In December 2020, the Amodei siblings officially left OpenAI and co-founded Anthropic.

Jack Clark, Sam McCandlish, Tom Brown, Chris Olah, and Jared Kaplan also participated in the company's founding, with Ben Mann later recognized as another co-founder.

Anthropic initially started as a small research institution focused on AI safety. Subsequently, the company gradually expanded to thousands of employees and grew to become one of OpenAI's most significant competitors.

Daniela Amodei now serves as president and chair of the board at Anthropic, overseeing most of the research, engineering, product, and business operations. Holden Karnofsky, who previously served as a member of OpenAI's nonprofit board, later joined Anthropic.

From simulating managers with stuffed animals to growing into an AI giant with thousands of employees, this story also witnesses Anthropic's journey from inception to maturity.

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.