Raw LLM Responses

It’s the same thing with testing for a rare illness, with some racial bias thrown in. In short, if you want to detect something that is rare with any accuracy you need an extremely good test. See the base rate fallacy. Most people don’t view facial recognition in this way and just think a hit is good cause for investigation. Further, a lot of facial recognition algorithms perform way more poorly on people of color, amplifying this problem. It is quite difficult to remove biases and issues from AI tech, since any racial bias in the training set is learned by the algorithm. edit: since the wikipedia article has a (poorly written) [terrorist example](https://en.wikipedia.org/wiki/Base_rate_fallacy#Example_3:_Terrorist_identification) you could also view this from the perspective of the fourth amendment: is it acceptable to be subject to a search where the main cause is a test that is less than 1% likely to be right?