Microsoft CEO: Model developers complaining about distillation is hypocritical Microsoft CEO Satya Nadella subtly criticized the way AI labs train their models. In a post on X last Sunday, Nadella stated that model developers complaining about distillation techniques is hypocritical. Model distillation refers to the process of using the outputs of a more powerful model to train a less capable one. 'While model providers have the legitimate right to use publicly available data to train their models, leading to great innovation, I find it ironic that the current situation is one where they advocate for this right on one hand, yet impose strict restrictions on model distillation on the other, while reserving the right to learn from customer usage and interaction data,' Nadella said. He also added that if knowledge can only flow in one direction, the party controlling the learning infrastructure will take all the profits, while the creators of knowledge won't be able to share in the corresponding benefits. Damn, why does that last sentence feel so familiar? Creators on X seem to be apologizing for this every single day. #Relatable