OpenAI’s Defense Against The Dark Arts: A Lesson in Power Dynamics
In a recent turn of events at OpenAI, the renowned artificial intelligence (AI) organization, power dynamics took center stage. The ouster and subsequent restoration of Sam Altman as CEO resembled a captivating soap opera that captivated the attention of people across America. However, for one Yale School of Management professor, it was a valuable gift that highlighted the significance of power and influence in organizations.
The board of directors at OpenAI held formal authority, but it was evident that Sam Altman wielded real power. This power play enabled him to solidify his control over one of the world’s most influential startups. It raised the critical question of how we can ensure successful governance of AI’s development, necessitating a deeper understanding of power dynamics and the empowerment of the key players responsible for safeguarding its progress.
While businesses are often seen as economic entities or social constructs shaped by culture and relationships, we tend to overlook their inherently political nature. Every organization is influenced by politics—the constant struggle for power. Although politics may make us uncomfortable, it is essential to recognize that even if we are not interested in power, power is interested in us. Understanding power dynamics becomes crucial because individuals with fewer ethical restraints will exploit it against those who fail to comprehend its nuances. It is in this context that the professor aptly renames their class Defense Against the Dark Arts instead of the formal title Power and Influence.
Formal authority stems from an organization’s written rules and can be seen in the board’s ability to hire and fire CEOs. OpenAI’s board possessed this formal authority, allowing them to remove Altman from his position. However, formal authority pales in comparison to the power that an individual can possess. Power, as defined by Battilana and Casciaro, is the ability to influence another person’s behavior by controlling a resource they value. In OpenAI’s case, highly skilled software engineers, immense computational power (particularly GPUs), and substantial capital are the critical resources at stake. Without any of these resources, OpenAI would undoubtedly fail.
Who controls these vital resources? OpenAI has approximately 770 employees, with almost 740 signing a petition threatening to leave the company if Altman was not reinstated. Only a few companies possess the computational power required to meet OpenAI’s demands, and even fewer have the financial means to support its endeavors. Notably, Microsoft, with its extensive investment of over $10 billion in OpenAI, controls these critical resources. The company’s influence over OpenAI became apparent when Microsoft CEO Satya Nadella learned about Altman’s ouster just a minute before the public announcement.
The board’s power play proved disastrous, as any student who proposed such a plan in the professor’s class would undoubtedly fail. Altman swiftly restored his authority in less than a week, and Microsoft now holds a board observer seat after previously having no representation. OpenAI’s ownership structure, with a nonprofit entity dedicated to AI development owning the for-profit research arm, was meant to prioritize safety over commercial interests. Yet, when confronted with the full power of a trillion-dollar corporation and a CEO described as built for a knife fight in the technocapitalist mud, these structural safeguards proved feeble.
This episode at OpenAI highlights a crucial lesson: neither clever structures nor well-intentioned individuals can overcome the pressures of commercial interests. If the development of AI is to address the concerns raised by prominent scientists and prioritize caution, market forces will inevitably prevail over idealistic pleas. Only power can effectively counter power, and the institution with the strength to push back against a multi-trillion dollar corporation is the United States government.
In the words of Georges Clemenceau, the French prime minister who led his country to victory in World War I, war is too important to be left to the generals. Similarly, AI, a technology of immense consequence, cannot be left solely in the hands of capitalists or scientists. It is high time for the government to establish a regulatory framework for AI before it becomes too late to mitigate potential risks.
As the dynamics of power continue to shape organizations and industries, OpenAI’s journey serves as a valuable lesson that highlights the need for a balanced approach. The delicate balance between power, influence, and governance must be struck to ensure responsible and ethical development in the realm of artificial intelligence.
