Home Cyberlaw Ars Technica Will AI fix prior authorization—or make it worse?