During the ongoing OpenAI trial, Elon Musk himself has created one more controversy. In the federal court hearing, Musk agreed with the fact that his AI venture called xAI has drawn on models developed by OpenAI to refine its own system. This is a renowned, controversial practice in the industry and is called model distillation.
For those unaware, model distillation refers to a technique where a larger or more advanced AI system is used to train a smaller model, effectively transferring knowledge and improving the performance. While the approach is widely used within organisations to optimise their own models, it has also become a point of debate when applied across competing companies.
During the questioning, Musk described distillation as using one AI model to train another. When asked directly whether xAI had used OpenAI’s models in this way, he skirted a clear confirmation, stating that such practices are common in the industry. Pressed further, he indicated that the answer was partly yes, adding that companies often rely on other AI systems to validate and improve the accuracy.
While it may not sound like any harm, it is unfair. Companies, including Anthropic and Google, have previously warned that distillation can, in some cases, be used to replicate capabilities without the same level of investment, raising questions around intellectual property and fair use. But still, it remains a widely adopted method to improve efficiency and reduce costs. And now, after the confirmation, this will become a bigger debate for the industry.
For the unversed, the ongoing trial was filed by Elon Musk against OpenAI, CEO Sam Altman, and President Greg Brockman. Musk alleged the company abandoned its founding non-profit mission to develop AI for humanity and instead became a profit-driven commercial entity through its multi-billion dollar partnership with Microsoft.