I would change your question from “does it have any real benefits” to “does it have a practical effect on the model” Benefits to me are clear: giving a develop

I would change your question from “does it have any real benefits” to “does it have a practical effect on the model” Benefits to me are clear: giving a developer choice over how their source code is used with for-profit, opaque, next generation ML models. But yes, drop in the ocean in terms of the full data set. But that shouldn’t be an excuse to remove user choice.