The risk from AI should not be conceptualized in terms of sentience or superintelligence