I’ve heard rumors that OpenAI is gearing up to a fully for-profit model, a departure from its previous non-profit structure. This has sparked considerable debate, with many expressing disapproval and even anger. However, I’d like to present a potentially unpopular opinion: I support OpenAI’s move towards becoming a for-profit entity.

Sustainability and the Bitter Lesson

Firstly, and perhaps most obviously, OpenAI needs to sustain itself. Their initial shift to a capped-profit model stemmed from the realization that scaling is paramount in deep learning, a concept known as the “bitter lesson.” This lesson emphasizes that scaling up models is the most crucial factor in advancing machine learning.

To achieve this scaling, substantial capital is required. Ensuring a continuous flow of capital is vital for OpenAI’s research. As a non-profit, they would be reliant on donors, potentially compromising their independence due to donor incentives.

Generating their own revenue through a for-profit model offers them greater autonomy and a more sustainable path forward. It’s a win for them in terms of financial stability.

Alignment with OpenAI’s Charter

Secondly, I believe this move aligns better with OpenAI’s long-term charter. Their stated goal is to benefit humanity by creating Artificial General Intelligence (AGI). This is a bold ambition, but realistically, such a powerful system won’t emerge in a vacuum.

Progress towards AGI will likely be a gradual, continuous slope rather than a sudden quantum leap. In the interim, if OpenAI develops meaningful systems that can provide value to the public, they should make them available, even if it means selling them.

The current large language models and voice models offered by OpenAI via subscription are not mere gimmicks. They offer real value. Providing these tools to those who can benefit from them, in exchange for payment, represents an efficient exchange of value for capital.

The Openness Debate

Finally, there’s the criticism that OpenAI isn’t truly “open.” I agree with this sentiment, but I believe they have limited choices in this regard. I don’t subscribe to the notion that they withhold models solely for safety reasons; I believe it’s primarily driven by competitiveness.

As mentioned earlier, securing capital is crucial for funding their research. Open-sourcing their models without restrictions would erode their competitive edge, potentially allowing other entities to surpass them. This realistic constraint, in my view, is the primary reason behind their closed-source approach.

Uncertainties and the Path Forward

While I’ve presented these arguments in favor of OpenAI’s for-profit transition, it’s undeniable that a for-profit company’s primary objective is to generate revenue and increase capital. Whether OpenAI can maintain its benevolent goals under this structure remains to be seen.

Despite these uncertainties, from OpenAI’s perspective, I believe this is the most logical and potentially beneficial step they can take. It offers them a path towards financial sustainability, continued research, and the potential to deliver valuable AI systems to the world.