From bots to self-drive cars, technology is not always all it’s cracked up to be, says Ian Taylor
Two recent ‘developments’ in technology are worth noting as they may, in different ways, prove of major significance.
One is a report by US cybersecurity firm Distil Networks which suggests that close to half (44%) of airlines’ website traffic is driven by bots.
That is double the average rate (22%) across all sectors and second only to that suffered by gambling sites (54%).
This hints at the size of the technological task required for network carriers such as British Airways, Lufthansa and Air France-KLM to switch from the current system of third-party distribution through GDSs to the world of personalised retailing promised by Iata’s New Distribution Capability (NDC).
Travel Weekly US reported the Distil Networks study, which points out the bots plaguing airline sites are engaged in web scraping, data harvesting and digital ad fraud.
Distil Networks’ senior director for security research Anna Westelius said the problem “is huge because there are a lot of bots holding airline seats for specific flights”.
These seats may be resold elsewhere – an issue Ryanair rages about repeatedly while bringing court cases against perpetrators.
I have not seen the study, but the report suggests it analysed “hundreds of billions of bad bot requests over thousands of domains” and identified the source of the attacks as primarily from “unauthorised online travel agencies, competitors, price aggregators and metasearch sites”.
Westelius noted this is not just a problem for the airlines: “It’s also a consumer problem because the prices increase.”
Now consider this in the context of NDC, which is expected to increase the volume of fare searches by a factor of between 10 and 100 – at least that is what Amadeus senior vice-president for travel channels Decius Valmorbida suggested at travel trade show ITB in Berlin in March.
Valmorbida reported Amadeus currently processes 20,000 fare searches and transactions per second and said: “With NDC, we expect volumes to multiply by 10 or 100.”
If the development of Amazon-style retailing through NDC connections with third parties seems slow among the network carriers, this is the reason.
The airlines want more control of the process – hence the imposition of surcharges on GDS bookings by Lufthansa, BA, Iberia and now Air France-KLM. But the carriers’ technology is barely up to handling the volume of fare requests.
Tesla crash
The second development was news of a fatal Tesla Model X car crash in California on March 23 which prompted shares in the self-drive vehicle manufacturer to plummet.
It led one leading analyst to query the “fallacy of automation” and another to “question Tesla’s promise that the current hardware will provide full self-driving capability”.
The authorities are investigating. Tesla had yet to reveal whether the car was on autopilot as this was written, but it issued an early statement defending the vehicle’s driver-assist feature.
The US National Transportation Safety Board is already investigating a fatal accident involving an Uber self-drive Volvo in Arizona earlier in March.
That crash saw a 49-year old woman killed while crossing the street pushing a bicycle after the self-driving system failed to slow or swerve the vehicle to avoid ploughing into her.
The person in the car at the time was reportedly watching a Harry Potter movie.
So in a matter of days the claims of the autonomous-vehicle lobby to be the harbinger of safer roads have been rendered hollow.
The only previous recorded death involving a self-drive car, in 2016, also involved a Tesla when one of its cars hit a truck.
Improved safety would be great – car accidents kill more than one million people a year around the world.
The US records an average 1.25 deaths a year on the road per 100 million vehicle miles. The multi-billion-dollar self-drive car industry set its sights on beating that by aiming to clock up 100 million death-free miles in trials.
Unfortunately, its record does not come close.
Uber reports its self-driving vehicles have driven “more than three million autonomous miles”. So the current Uber record is 30 times worse than the average for vehicles with a person at the wheel,
Tesla claims its vehicles have clocked up 130 million miles on autopilot, leaving it with a death rate of one per 65 million miles.
However, this is with the odds of avoiding accidents stacked in the self-drive sector’s favour.
Bryant Walker Smith of the University of South Carolina law school told the Financial Times: “The fact that this [Uber fatality] happened well in advance of 100 million miles does not tell us anything statistically. But it is early in light of everything these systems have going for them.”
He pointed out autonomous cars are well maintained and closely supervised, saying: “That should stack the deck in favour of these systems.”
Accident rates
In fact, autonomous cars have been involved in dozens of non-fatal accidents.
Almost 60 have been reported in California since 2014, where there are not yet 400 self-drive cars on the roads. (There are more in Arizona where the pedestrian was killed.) That is an accident rate above one per seven vehicles.
By way of comparison, there are 26 million cars in Britain and in 2016 these were involved in road accidents causing 155,500 minor injuries. If we assume just one minor injury per accident, that is about one per 170 vehicles.
Potential consumers have so far shown limited interest in self-drive vehicles. A survey of 1,500 US drivers last August found more than half (55%) “unlikely” to purchase an autonomous vehicle and 71% unlikely or unsure.
A more recent survey of 6,000 adults in the UK, Germany, France, the US and Canada on behalf of Travelzoo found only one in four said they would be happy to get in a driverless vehicle.
So what is driving the self-drive car race if not safety and not consumer demand?
In Uber’s case, removing the drivers from its cars (taxis) offers a pathway to profits. This is business losing up to $1.5 billion per quarter, remember.
For Google (Alphabet) and its Waymo car business, the technology offers the prospect of drivers being able to remove hands from wheels and minds from roads to play with their phones.
The industry will no doubt blame errant human behaviour for the deaths. But there is no way of removing human drivers from other vehicles in the foreseeable future, and no way to remove pedestrians or cyclists at all.
If self-drive cars can’t hack being in the same road space as people, there is no space for them.
More: Travellers doubt safety of driverless cars, pilot-less planes