Raw LLM Responses

Regarding predictive algorithms: Whatever the prediction, the algorithm is based on the percentage of those who are predicted to display the targeted behavior in the future. If the designers of the algorithm believe that they are able to identify persons who will display a targeted behavior in adolescence through adulthood, based on the existence of no house numbers on their homes or any of the other predictive identifiers of later criminal or anti-social or community unapproved activities, can the designers document that 100% of the time the identified child or adolescent will become an adult criminal. In medical testing and the behavioral sciences, there are no predictors that are 100% accurate. In making predictions, the best we can do is to identify the probability of failure of the predictors. In much of the social science research, the standards for validating an intervention accepts no more than 5% (p = .05) of the predictions being failures. In some cases, acceptable probability may be set at .01, .005, or an even more stringent probability for failure of the predictor. The probability of failure for algorithms of .05 means that one in twenty will be unlikely to meet the predicted outcome of criminal behavior beyond childhood. In medicine, the guiding principle is that a new drug or medical protocol must do no harm which is unrealistic. Many drugs or protocols go through development to the point of looking positive before human subjects testing begins. When drug or protocol testing on human beings begins and failure of expected outcomes occurs at an unacceptable level in the form of allergic reactions and other outcomes including death. The allowed unacceptable level varies with the severity of the consequences. If test administrations of a drug is followed by 996 being cured of the targeted condition, the testing program has beat the the predetermined .005 standard. If 20 or 3 of 1000 die, the drug has harmful consequences more severe than should be acceptable. What kind consequences should should be tolerated in a predictive policing program? What about the father who moved to Pasco County to pursue a law degree and has been damaged so severely that he has given up his dream? Are there children or adolescents and/or other family members who have been harmed in ways equally or more significant than those mentioned in the article? This program has a scary track record and potential for producing shotgun harm to individuals in ways that have been only partly identified. Do no harm.