Peter Norton - Autonorama - The Illusory Promise of High-Tech Driving

Any sufficiently advanced technology is indistinguishable from magic.

Arthur C. Clarke

Somedayand it may come surprisingly soona car maker will introduce a radically advanced automobile, and cash in on the giant market of tomorrow. So an American electronics firm seeks to entice automobile manufacturers interest in its products. The advertisement depicts a high-tech future: Four smiling occupants of a driverless car travel a high-speed, zero-congestion highway, enjoying one anothers company in comfort and safety. They face each other; none pays any attention to the road. Their geographic setting is nondescript; what matters are the fast vehicles and their relaxed occupants, the delay-free roadway, the unseen electronics that purportedly make this

Such promises of technofuturistic driving utopias, depicted vividly and with claims of credibility, have again been ubiquitous in the media and wherever tech companies and carmakers meet. They assure us of a future in which traffic congestion never slows vehicles down, and cars never collide with anything. These promises, in turn, become reasons not to commit too much money or attention to modes of transportation that are already safer today, and that are also spatially efficient, more sustainable, and much cheaper.

By promising perfection, the promoters of technofuturistic visions make ordinary sufficiency bland by comparison. According to the author of a blurb on the back of a 2018 book celebrating our driverless future: This book should be required reading for every city planner and for every taxpayer fed-up with wasted transit dollars.

Though diverse, the visions of high-tech driving share a common claim: with sensor data, state-of-the-art hardware, machine learning, and digital networking, onboard computers in every car will drive for us, better than we can, and sooner than we think. Despite the extraordinary technological developments of the last twenty years, however, the practical possibility of widespread automatic driving remains elusive. High-tech solutions, always just over the horizon, are supposed to offer the anticipated deliverance. The lack, however, lies not in technology but in the aspiration itself. Meanwhile the supposed solutions, in promising an eventual end to all our afflictions, divert us from transport sufficiency: an unspectacular state in which everyone can meet their practical needs.

The governing assumption is that a car-dependent future city that is efficient, safe, sustainable, and equitable is practically possible, or desirable enough to be pursued at any cost, to the neglect of less utopian but more feasible alternatives. When pressed, promoters of such futures point out the obvious deficiencies of the status quo, as if the universe of our choices is limited to just two possibilities: status quo car dependency or futuristic car dependency. We may hear that people prefer to drive. But in settings that offer no good alternatives to driving, we cant say what people prefer. More often, however, we hear that the technology, like the weather, is coming; for those of us not developing and selling it, our task is only to forecast and prepare for the inevitable. More often than not, in the advocacy of machine autonomy there is an implicit denial of human autonomy.

There are no autonomous vehicles

Most of the autonomous cars we are presented with look a lot like conventional cars. Those that are coming are supposed to be electric, but an electric car can be human driven; the high-tech enthusiasm is about the cars supposed autonomy. The implication is that cars problems are not due to their spatial demands, their low passenger capacity, or their weight (and consequent energy requirements) but to their human drivers. In this case, then, the remedy is a driverless car, or rather a car that drives itself: an autonomous vehicle.

Yet the term autonomous vehicle (AV) is paradoxical. Engineers have specialized definitions for autonomous systems, but among wider audiences, something autonomous has a will of its own. AV promoters seem to welcome the association, perhaps because it suggests that the AV is in control, unimpaired by human deficiencies. A vehicle that is in fully autonomous mode accelerates, brakes, and steers without direct human supervision, but only in response to the dictates of sensor data and the program that processes them. An AV is controlled by its program just as a conventional car is controlled by its human driver. The program has been automatically trained on vast data throughmachine learning, but human beings decide how the car will respond to its environment. Will the AV be risk averse, and drive so cautiously it frustrates its human occupants? Will it take chances, thereby offering its passengers a better ride? Will it yield to all pedestriansor will it first honk at those who are not in a crosswalk? Will it comply with all the rules of the road so scrupulously that it annoys the human drivers behind it? Will it apply its machine learning capacities to determine how much faster than the speed limit it can go without risking a penalty? The car does not decide. Decisions like these have already been madeby human beings.