How Things Fall Apart
Society falls in love with ideas all the time. Ideas that will never happen or at least no time soon. Flying cars, video phones, 4K Ultra HD, and driverless cars. It is not that humans are incapable of creating the base technology. It is more a matter that humans are the finicky sort. Just because we can, doesn’t mean we want to. Or sometimes it just isn’t worth paying for.
Plenty of technology has been available for a good long time. Jet packs, super-sonic planes, moon bases, deep ocean colonies, and video phones. All of these could have been here decades ago, if only technology were enough. In all honesty, self-driving cars need little more than some good data and some layered heuristics to do the vast majority of the driving. It is a two-dimensional game space with some fairly structured rules… Data Science is really just an optimization play.
Well, at least until this happens…
Seems harmless enough, doesn’t it? Welcome to the Festival at Bel Air where the laws of driving cease to apply. Seriously, if you want a Turing test for driverless cars, bring them here. If they can navigate this intersection, they might have a chance. Although, Bel Air is hardly New York City at rush hour, which honestly pales against intersections across Europe.
Don’t get me wrong. If everyone followed the rules, things would be fine. But how will Google’s (or Uber’s, or whoever’s) A.I. handle humans being human? Things are liable to get really inconvenient or really dangerous. Of course, I could be completely wrong… only that hardly matters.
Even if the technology can account for human behavior, things only have to go really bad once or twice. Remember the space program? Not the Space-X one, the one that featured real people going to real places (like the moon) for real reasons. How about the Concorde? Or Uber in Arizona? A few bad things will ruin a good one in a hurry. Only our AI overlords need not have such bad luck, just a little will be a problem.
If Google cars start crashing, things are going to get interesting in a hurry. Today, drivers in the US sue other drivers. It is a complicated affair with numerous intermediaries, across an array of jurisdictions, and concepts like ‘no fault’ and ‘full coverage’. But at the core, it is distributed risk.
When Google, Uber, Apple, or whoever’s AI is driving the car — who will drivers sue? The car insurance industry is going to have to change. What about when your Google car hits a pothole or a deer? Who is liable now? And what for? Risk matters.
After the Uber Arizona incident, much was made of the “attendant” being otherwise engaged. Does anyone believe that driverless cars are going to have fully engaged passengers? Perhaps they think that the car will simply give control back in these areas of “lawless” driving? Or that each car “attendant” will be held liable as the final line of defense? Sorry, can is great… will is another story.
Longer term, do you want the Google AI to hand control to occupant who only drives during exceptionally confusing times? How good will they be? How will anyone ever get a license? What will driver’s licences need to become?
Does any of this sound convenient? Markets will try anything that doesn’t fly in the face of human interest, but things only gain traction when they are convenient, time saving, and money saving. If I have to baby sit and audit my AI chaffeur, how am I saving time? If I have to consider new licensing and insurance laws, how is this easy or convenient? AI will need to be better drivers so that perhaps I can save money… although insurance companies are likely to fight that.
I haven’t begun to consider car prices. With CAFE standards, safety standards, tariff wars, and a myriad of other woes — these things aren’t getting any cheaper. Everything else is — TVs and computers, smartphones and even helicopter rides. How is adding a super computer and a myriad of cameras and sensors to the car going to make that any better. My smart phone is out-dated after two years, will my $45K smart car last that long? It will be a decade before we could even begin to know…
Once upon a time, trains ruled the world. Remember the robber barons? Now we have Amtrak… they are still in business right? Remember the Wright Brothers? Of course you do, today they sit right along side Boeing… ummm, Southwest as… actually they didn’t even outlast Eastern. If it weren’t for the billions of bailout dollars in 2002, the list of failed airlines would be far larger. Point being — technology often changes the world, but not before bankrupting a lot of smart people and never as quickly as anyone imagines. Isn’t time for Popular Mechanics to try again?
Oh right, looks like PopSci beat them to it. Well, actually I think they were just poking fun. Even this article is now a decade old. Maybe flying cars will have a shot at self-flying? There is no legacy infrastructure to stand in the way. In the end, it won’t be up to our AI Overlords. It will be all about us silly humans.
Thanks for reading!