I agree -- we are capable of avoiding life/death scenarios and it is precisely this technology that can get us there. We may see more nuanced ethical dilemmas, however. AI driving could eliminate the need for traffic control elements like stop lights/signs, so do we tune the algorithm to optimize individual vehicle metrics (e.g. fastest time, lowest fuel consumption, lowest risk) or societal metrics (e.g. lowest combined drive time, lowest congestion, lowest pollution)? Do we allow people to customize their algorithms or will they have to be standard to cooperate? Will jailbroken/rooted vehicles endanger others (they will surely exist)?