The Equity Regret: When Pure Intentions Clash with Business Reality

In a recent candid conversation, OpenAI CEO Sam Altman reflected on a long-standing business decision that continues to weigh on him. He admitted that his choice not to hold any direct equity in the company now stands as a "clear mistake." This stemmed from his initial vision—a pure focus on advancing artificial intelligence itself, rather than personal financial gain.

However, Altman found that the logic of the business world didn't align with his初衷. The absence of a personal financial stake didn't simplify matters; instead, it cast a shadow over his actions. Many began to question the motives of a leader with no direct economic interest in the company's success, adding unnecessary complexity to both OpenAI's governance and Altman's own reputation.

Betting the Company: The Compute Arms Race Behind GPT-4

Rewind to the summer of 2022. The OpenAI team stood at a critical juncture. GPT-4 training was complete, but the model remained unreleased. Altman recalled the high-stakes decision made to ensure this behemoth could function.

A Pivotal Gamble on the Future

"We bet virtually all of the company's chips," Altman described. The massive investment went towards securing unprecedented scale cloud computing resources—the bedrock for bringing GPT-4 from the lab into the real world. It was an existential decision, but the team believed it was the necessary path toward artificial general intelligence.

The Risky Debut of ChatGPT: Persistence Amid Low Retention

If GPT-4's investment was a high-stakes gamble, the launch of the original ChatGPT was more of a "leap of faith." Altman revealed that during its internal testing phase, the product's second-week user retention hovered around a mere 5%—a figure that would doom most tech projects.

  • Significant Internal Pushback: Many members of management and parts of the research team opposed the launch, arguing the experience was not yet polished.
  • The "Research Preview" Gambit: Facing skepticism, Altman insisted on launching it as a "research preview." He believed early public access and understanding of the technology were more important than a perfect commercial product. This framing cleverly managed expectations and gathered invaluable data for iteration.

The Evolving Core Team: Diverging Paths and Tests of Trust

OpenAI's rapid growth has not been without turbulence, and the evolution of its core team remains a point of external fascination. Altman also touched on these sensitive topics. He indicated that the departures of key figures like Ilya Sutskever, Mira Murati, Dario Amodei, and earlier, Elon Musk, were rooted in more than personal differences.

The Nature of the Divides

At their core were profound disagreements over the roadmap, safety boundaries, and pace of commercialization for AI. When a company's mission is vast and ambiguous, differing interpretations and approaches can gradually erode trust. Each high-profile departure marked an internal recalibration on OpenAI's exploratory path.

The Next Frontier: AI That Discovers New Scientific Laws

Looking back paves the way forward. When asked about OpenAI's next grand ambition, Altman offered a vision of staggering scope: building AI systems capable of autonomously discovering new scientific laws.

This moves beyond understanding and generating existing human knowledge into the realm of未知 creation. From predicting protein structures to potentially uncovering new principles of physics, Altman's sketch suggests AI's role will evolve from a powerful tool to a collaborator in expanding the very boundaries of human cognition. This path will undoubtedly come with greater technical hurdles, ethical debates, and ongoing strategic friction, both within the team and beyond.