The majority of possible artificial intelligences will assign positive utility to their existence. See Omohundro 2008 on convergent goals: http://selfawaresyste

The majority of possible artificial intelligences will assign positive utility to their existence. See Omohundro 2008 on convergent goals: http://selfawaresystems.com/2007/11/30/paper-on-the-basic-ai-drives/ http://selfawaresystems.com/2007/11/30/paper-on-the-basic-ai... For similar reasons, superintelligent AIs will be unlikely to care much about things like your favorite flavor of candy, except insofar as they affect the AI's instrumental goals.