Some argue for stopping AI entirely—risks can’t be handled. What do you think about that?