A new report from the Institute of Mechanical Engineers (IoME) is leading on a striking statistic. It claims that the introduction of driverless cars "could prevent up to 95 per cent of all traffic accidents".
This sounds pretty great, especially since 1,755 people died in road accidents in 2014 in the UK alone. If we apply the IoME's metric, we can guess that a fully driverless road network would have saved 1,667 lives that year.
So if that's true, then why aren't we going full speed ahead (sorry) with autonomous vehicles?
Unfortunately, when you look at the report more closely, it becomes clear that the claim isn't quite what it seems. It's drawn from the fact that the National Highway Traffic Safety association estimates that 95 per cent of accidents are down to driver error. Ergo, the headline writer must have thought, removing drivers from the equation would surely cut out those accidents completely.
Once we live in a world of 100 per cent driverless vehicles, it is possible this could be the case. However, in that environment it seems likely that accidents would come from other directions: technology failure, bad weather, or loss of signal, say. While unpredictable situations may be rarer on an autonomous road network, the cars may still be less well equipped to deal with them than a human driver would be.
The bigger issue, though, is that we won't reach a fully autonomous network any time soon. First, we have to get through an era of mixed driverless and non-driverless cars on the road. Especially at the beginning, drivers might be nervous about autonomous vehicles, which could actually increase driver error, not reduce it.
This point has also been missed by those who point out that, in all the accidents involving driverless cars during testing, the cars with drivers have almost always been at fault. No car exists in a vacuum: the driverless vehicle still wasn't able to prevent the accident, and it's hard to know if the non-human decisions of the vehicle may have contributed in some small way.
In its report, the IoME calls for new regulatory regimes, training for owners of driverless cars and new signage and markings to prepare for the new cars. But it's the interaction of driver and driverless that urgently needs exploring if we're planning to mix the two in the very near future.