The ethics of self-driving cars (revisited)

The discussion about the ethical programming of autonomous vehicles continues, and it seems like we tie ourselves in ever more Gordian knots over it (I have already looked at this issue in some detail).

Another article in today's Globe and Mail Drive annoyed me a bit, though, because of the line of thinking of some participants in the debate (mainly car manufacturers) who believe that we are putting unfair expectations on self-driving vehicles. As an example: "I don't remember when I took my driver's license test that this [in reference to the need for autonomous cars to decide which victim to pick in the case of a potential accident] was one of the questions". Others point out that the concerns being debated over the ethics of future self-driving cars does not seem to extend to existing vehicles and their current capabilities.

It seems to me that this is an entirely moot point. The whole issue with autonomous vehicles is that they do not have a human in control, a human with decades of accumulated life experiences and millennia of social, psychological and evolutionary development. We have to try and teach these machines all of that in just a few years. If we are putting life and death decisions in the hands of technological mechanisms, then we need to do the very best we can. This is not unfair expectation, it is merely due diligence, and "doing the right thing". And the time to do this is now, at the very beginning, before lives are at stake (oops, too late!)

Another aspect of this debate which I had not thought much about is the idea that autonomous cars need to communicate with pedestrians. We need the equivalent of a pedestrian's ability to make eye contact with a driver to have confidence of what they are intending, or the simple nod or light flash of a driver to a pedestrian or another driver, indicating their awareness of a situation and their intentions. There are many such non-verbal communications in use daily on our roads, many of which we are not even aware of ourselves.

How, then, is a driverless car to make such nods and eye contacts? Various ideas have been suggested, including flashing text (language barriers?) and symbols (low recognition rate) and a system of flashing lights (need for a national/global standard and pedestrian education). What I found interesting, and a bit depressing, though, was the research using fake driverless cars which showed just how little eye contact there really is these days. Only 19% of pedestrians ever looked up at the cars around them (19%!), most being more intent on their phones and conversations. So, most people would not even be aware that they are facing an autonomous vehicle, and any amount of winking and flashing the cars may be engaged in is likely to be pretty much wasted.

You can kind of see why vehicle manufacturers might be a bit jaundiced in their outlook...