If statistical proof of the safety of automated vehicle driverless mode is impractical, what then?
An interesting statistical analysis by RAND shows that proving the hypothesis that fully automated (driverless) mode is safer than human driving is likely to be well beyond what is likely to be possible. So is our society prepared to accept less than a 95% confidence level for that statement, can simulated driving genuinely substitute in the assessment, or should we wait a very long time to prove beyond reasonable doubt that something generally accepted to be true actually is?
Read the report here.