Guidelines for AI Quality Control

Some models will update themselves and learn, like, on the fly, they call it online learning. Those probably need daily assessments to make sure that the outputs they're making are sensible and ethical and unbiased. As opposed to more static model predicting, you know, risk of cardiac events, that doesn't necessarily change overnight. That could take a little bit less oversight, but absolutely, you need to have some human element overseeing this model to make sure its doing what you intended it to do.