My column in The Times:
The Civil Aviation Authority is concerned that pilots are becoming too reliant on automation and are increasingly out of practice in what to do when the autopilot cannot cope. We now know that a fatal Air France crash in the Atlantic in 2009 was caused by confused co-pilots reacting wrongly when the autopilot disengaged during turbulence. They put the nose of the plane up instead of down.
But there is another way to see that incident: the pilot was asleep at the time, having spent his time in Rio sightseeing with his girlfriend instead of sleeping. When roused as the plane stalled, he woke slowly and reacted too groggily to correct the co-pilots’ mistakes. Human frailty crashed the plane, not mistakes of automation.
Human error, or sabotage, also seems most likely (though we cannot yet be sure) to have disabled and diverted the Malaysian Airlines jet that vanished over the Indian Ocean in March. Human action certainly caused 9/11. For every occasion on which a Chesley Sullenberger brilliantly and heroically landed a plane on the Hudson River after a flock of geese went into the engines, there have been many more where people caused catastrophe. Human error is the largest cause of crashes in the sky, as it is on the ground.
That is, I suggest, why we will embrace the inevitability of pilotless aeroplanes at some point in the not so distant future. Already, automated systems are better at landing planes than pilots, even on to aircraft carriers: they react quicker. Drones are crashing less often when allowed to land themselves rather than be guided in by ground-based pilots. Even Hudson River heroism could possibly be automated. I confess I am probably an outlier here and that most people will be horrified by the prospect of boarding pilotless planes for a while yet. But I think they will come round.
Driverless ground transport will help to assuage our fears. I took a driverless train between terminals at Heathrow last week, and Transport for London has begun tendering for driverless Tube trains, to predictable fury from the unions. Prototype driverless cars are proving better and safer than anybody expected. It cannot be long before they seem preferable to an occasionally distracted, risk-taking, radio-playing or grandee-teasing taxi driver.
Google’s prototype self-driving cars have now covered more than 700,000 miles on public roads with only one accident — which happened when a human took the controls. They may be commercially available after 2017. Testing of self-driving cars will begin on British roads next month.
Getting out of a driverless car, after a restful journey working and reading, then telling it to park and come back when you need it, would bring the luxury of the chauffeured plutocrat within reach of ordinary people. Driverless lorries on the motorways could be confined to night-time operation, leaving the roads clear for cars in the day.
In the air, small drones are now commonplace and not just in the military. The “Matternet” is a plan to use them to supply the needs of remote areas with few roads in poor countries, leapfrogging poor infrastructure as mobile phones leapfrogged the lack of landlines. Once drones can refuel each other in the air, they should quickly take over (for instance) searches of the ocean when planes or boats are lost — so as to put fewer lives at risk.
The next step would be that cargo planes would fly without human beings aboard. The sticking point will be air-traffic control’s reluctance to sanction such planes landing at airports in built-up areas. At the moment, drones and piloted aircraft are kept apart in separate zones. If you live under a flight path it is comforting to know that the planes overhead are piloted by people with every incentive to land safely: with “skin in the game”. The existence of a “ground pilot” who can take control of a plane from the ground, as drone operators can do now, would be of little comfort to such people, let alone topassengers on a plane.
But pilots’ wages and training costs are one of the highest contributors to the cost of flying, after fuel, and if pilotless planes can fly safely for years without passengers, objections to them carrying passengers will gradually fade. An ordinary aircraft is now regularly flying between Lancashire and Scotland with nobody at the controls(though there is a crew on board to take over if necessary). The offspring of a seven-company consortium called ASTRAEA, it uses radar, radio and visual sensors to detect and avoid hazards.
Are we approaching the era when it will be more reassuring to know that there is not a human being in the cockpit than to know that there is? We might find it comforting to know that the cockpit was wholly inaccessible to terrorists and that the machine within it had not spent the night drinking.
It is true, as the CAA has spotted, that we currently have an uncertain mixture of people and machines flying planes, with a danger that the former are getting out of practice and confused. But since accident rates are low and falling, there is no evidence that this partial automation has been a problem, or that going further towards full automation would not help.
Perhaps robotic surgery holds a lesson. Justin Cobb, a distinguished professor of orthopaedic surgery at Imperial College London, tells me that his engineers build into his experimental robots — which carve out, via keyholes, slots in your knee or hip bones of just the right size and shape to fit the necessary implants — what is little more than an illusion of control by the surgeon. The surgeon is allowed to move the tool about, but only within a certain boundary. Beyond that, the robot’s software prevents the tool straying.
So an automated aeroplane might allow the pilot to play with the joystick and the switches, but only within limits. Thus can the pilot retain what is left of his dignity and the passenger indulge what is left of his irrational fear of submitting his life to a machine. Imagine a future hijacker or suicidal pilot finding the controls of the plane refusing to obey orders. Like Hal in the film 2001, but in a good way: “No, Dave, I can’t let you crash this plane.”
So in practice, despite the cost, we will keep pilots around in the cabin even if there is not much for them to do, and surgeons in the operating theatre, farmers in the cabs of tractors, teachers in the classroom, lawyers in the courts, and columnists on newspapers.