Autonomous Vehicles
Associated Press, DECEMBER 21, 2025, Washington DC: Food riots are breaking out in major American inland cities as the ongoing logistics crisis enters its third week.
More than 80 per cent of the country’s logistics fleet remains off the roads after a disastrous software update shipped on 30 November. Only vehicles still under the control of human operators are currently available for food delivery …
Driverless vehicles remain one of the tech industry’s favourite futuristic scenarios, fuelling a daily run of announcements, partnerships and promises. Yet the scenario set out above is as reasonable as any of the industry’s more bullish forecasts.
Since 2018 began, the IT industry has watched, in horror, the slow-motion train wreck of Meltdown/Spectre vulnerabilities.
Why? Because these are hardware bugs and they’re harder to deal with than a slip in a C++ library.
More importantly, because different hardware vendors—Intel, their rival AMD, and high-profile Internet-of-Things supplier, ARM—used similar approaches to ‘speculative execution’, these are cross-platform hardware bugs that could potentially show up anywhere.
Does the idea of a largely-automated truck fleet being disabled by a buggy firmware update, still seem far-fetched?
Some kind of autopilot
For now, we aren’t too dependent on driverless technology.
We’re still waiting to see who gets there first: Tesla, Google, or the auto industry.
In a way, it’s odd that Elon Musk dubbed the Tesla autos’ self-driving capability “autopilot”, when even a car with minimal self-driving safety must do so much more than an aeroplane.
Apart from recent systems that can do everything but take off, most autopilots don’t do much at all, beyond adjustments to keep an aircraft on course, at the right altitude, and level. You certainly don’t need a computer for most autopilot work; the earliest technology is more than a century old.
Compare that to the constant attention required to keep a car safe for even a simple task like staying on a divided highway. What Tesla and others, like Volvo (with Uber), have achieved is far more than most of the autopilot technology in aircrafts around the world.
Good enough?
In isolation, driverless car safety might be the easiest problem to solve. Not because we’re even close to perfection, but paradoxically because perfection might be a misguided goal.
Late last year, think tank The RAND Corporation offered up this analysis of autonomous vehicle safety trade-offs.
It sounds cold-hearted to say, but if driverless cars are better than humans, the RAND Corp analysts argue that the results are good enough, albeit not perfect. Yes, people will die in accidents involving autonomous vehicles, but if more people would die in human-controlled cars, the trade-off results, overall, are beneficial.
The RAND Corp analysts’ cold equation is informed by the truly vast death toll on American roads: 35,000 plus in 2016. This is a death rate of 11.59 per 100,000. Australia’s rate in 2015 was 5.08 per 100,000 and Europe, by comparison, reported around 51 road deaths per million inhabitants.
That raises a steeper question about driverless cars: whose standard prevails? Reaching an acceptable level of safety in America is vastly easier than in Europe. And conversely, cars safe enough, on balance, to save lives in America would create carnage in Australia or Europe.
We’re not going to get there all in one disruptive leap. And if you ask experts at NASA, there’s a very dangerous overlap period we may have already entered.
Not autonomous enough
NASA has spent decades understanding automation risks in aircrafts, and last year told Scientific American that today’s cars “equal airplanes in 1983”. By that, NASA research psychologist, Stephen Casner, means we already understand the interaction between humans and semi-automated vehicle.
A fully-automated vehicle is not the problem. Issues arise when the car assumes most of the cognitive load, leaving a bored human unable to respond quickly in an emergency situation. Casner told Scientific American that the attention problem is more intense for a driver than any pilot (even fighter pilots): you must be ready to respond to anything, all the time.
Perhaps flying cars are a better solution after all. Provided air traffic control is scaled-up to cope with many more movements, autonomy in the air is much easier than in cars on the road.
The Strava debacle
An International Studies student at the Australian National University noticed that fitness app, Strava, had published data which mapped a bunch of military bases.
No matter how many times apps accidentally break privacy, there’s always another company ready to publish first, regret later.
There’s another important parallel between Strava’s mistake and the future data-gathering potential of autonomous vehicles: a lowest common denominator attitude to privacy is not good enough.
We need to make sure that auto-makers obey local privacy rules now rather than being dragged to the table after probable mass privacy breaches.
When that’s done, everyone can own autonomous cars and discover there’s a hardware bug common to all platforms …
About the author: Shara Evans is recognized as one of the world’s top female futurists. She’s a media commentator, strategy adviser, keynote speaker and thought leader, as well as the Founder and CEO of Market Clarity.
Leave A Comment
You must be logged in to post a comment.