The recent turmoil at OpenAI, marked by the firing of founder Sam Altman by the board, reflects deep-rooted disagreements and challenges within the organization. Altman and the board disagreed on AI development speed, business strategies, and AI risks. Also, Altman’s side projects, like raising funds for an AI chip startup, added to the tension. This upheaval not only highlighted fundamental conflicts within OpenAI but also led to the existential crisis of the company, a beacon of AI development. This blog explores three key areas of conflict that I think define OpenAI’s current predicament.

Non-Profit Goals vs. For-Profit Demands

The conflict between OpenAI’s non-profit mission and its for-profit operations lies in control and distribution of profits. OpenAI’s for-profit subsidiary, OpenAI Global, LLC, is fully controlled by the non-profit organization. While the for-profit subsidiary is permitted to make and distribute profit, it is still subject to the non-profit’s mission. This structure has raised questions about decision-making processes, board composition, and the influence of major investors. It has created some challenges and tensions within the organization.

Adam D’Angelo’s Role Raises Questions

Adam D’Angelo’s company, Poe, offers similar products to OpenAI. Poe is a general knowledge chatbot similar to OpenAI’s ChatGPT. It also allows users to create their own bots using language models like ChatGPT and other alternatives. Additionally, Poe has a subscription tier and a marketplace where custom bot creators can earn revenue from users. This puts Poe in direct competition with OpenAI’s ChatGPT and their GPTs product concept. To be honest, I don’t understand how Adam can maintain a role on OpenAI’s board while running a business that competes with OpenAI at the same time.

OpenAI and Microsoft: A Complex Partnership

The partnership between OpenAI and Microsoft is strange. It’s true that their current collaboration looks good: “OpenAI provides the best models, and MS provides the best systems,” as Satya Nadella said on OpenAI’s Dev Day. But, MS is providing almost the same service using OpenAI’s model through Azure’s cloud computing. They are not only partners, but also competitors. The partnership agreement says once OpenAI achieves its mission of reaching AGI, Microsoft may be out of the loop. For MS, obtaining OpenAI’s advanced models and team while disregarding the AGI restriction seems the best option. Now, with Sam and other key employees joining MS, I think MS is the biggest winner so far.

Conclusion: Lessons on Mission Alignment and Avoiding Conflicts

This situation at OpenAI teaches important lessons. Having a simple and clear mission is not enough; an organization’s operations should really match this mission. To avoid confusion and turmoil like we are seeing at OpenAI, organizations need to do their best to remove conflicts of interest in the team, structure, and operations.