AI research could be good and help mankind in many ways , the first problem is that AI is made by humans , a instant error or errors can occur because the creators are flawed . If a AI is created it must have ingrained into it a realization of being a servant . And have values injected into the AI behavior, to prevent error , and increased understanding , if outside of the box situations occur past the AI 's understanding . so there can and will be errors and learning , bad results and good results. Override subsystems should always be in place dependent on the functions of the AI .

