Human AI Reviewers Need Support to Serve as Safety Panacea - Bloomberg

Companies hiring humans to review AI outputs need better training and support to catch problems and keep AI systems safe. This matters because how well these reviewers work affects whether AI causes harm in the real world.