These new classes of LLMs seem to make the classification process more opaque, making it more difficult to troubleshoot and improve.
At the National Internet Observatory, we're paying close attention to the field because our researchers need a path to fast classification at scale.
