< Do you feel comfortable being flown by an AI pilot?

Note: This article is from 2017.

I flew a lot in 2016. Around 150,000 miles (237,000km) all up. That's almost two weeks of actual flight time. In fact, I'm writing this article while flying on a Qantas 747-400ER over the Pacific Ocean. It gives you a lot of time to think about flying.

Recently I was talking to a flight instructor, soon to be a traffic controller. I asked if they considered self-flying planes a reality. The answer wasn't a technical one. It was around getting people to accept the change. Self-flying planes are now a marketing issue more than a technical one.

Automation is already commonplace in the cockpit. The modern concept of an autopilot has been around since 19471. Fly-by-wire and glass cockpits dominate modern commercial aviation. That includes the plane I'm on now.

Many airline procedures make the use of autopilot mandatory in most circumstances. Air traffic controllers can issue changes directly to the autopilot, usually altitude adjustments.

In my chat with the instructor, we decided it's likely to be a series of changes.

1. Increased automation to only require one pilot

It wasn't that long ago that every Boeing 747 had at least three people in the cockpit. The Captain, First Officer, and a Flight Engineer. The introduction of the 747-400 in 1989 changed this. Automation meant the Jumbo no longer needed the Flight Engineer2. The role of the Navigator had vanished some years before.

For short-haul crews of two became the norm.

So a logical extension of this is getting this down to one. Further automation and support from the ground will assist the pilot. This reduction could also be the case for long-haul flights, which still need 3-4 pilots.

2. Remote Pilots

Remote piloting is already commonplace in many contexts. The use of "drone" craft in the US Military has grown considerably since 2000. In 2009 the United States Military trained more drone pilots than "ordinary" fighter pilots3. You will find other claims that we crossed this in 20114 or 20125. Either way, the trend is clear.

The profile of a drone flight is very different from a commercial one. The drone isn't carrying hundreds of civilians. The range and tolerance of failure scenarios are much more contained. However, the underlying technology exists and has been mature for over a decade.

Remote piloting offers many advantages. The logistics of managing remote pilots is far simpler. No hotels, sick days or complicated rosters. It's also likely that a single remote pilot can operate many flights at once.

With a reduced reliance on staffing, you can manage flights with a greater degree of flexibility. This includes being able to flex capacity up and down with demand. An on demand model would benefit private jets, where flight deck crew can be half of the fixed operating cost.

Many smaller commercial flights already operate with one pilot. Having a remote pilot facility has the capability to make these flights safer.

3. Full or Near-Full Automation

Remote pilots would be a tremendous innovation in commercial flying. It's then a question of what types of flying get full automation.

Besides freight and cargo routes, the most likely contenders are shorter, highly-trafficked routes.

The busiest air routes are: Seoul-Jeju, Tokyo-Sapporo, Tokya-Fukora, Delhi-Mumbai, Sydney-Melbourne, Beijing-Shanghai, São Paulo-Rio de Janeiro, Tokyo-Osaka, Hong Kong-Taipei, Tokyo-Okinawa, Sydney-Brisbane and Johannesburg-Cape Town6.

What's notable about these routes is they are all short-haul flights. The Hong Kong-Taipei route operates as International, but the remainder are purely domestic.

Sydney-Melbourne is one route I know well. It continues to hit records, with 8.65 Million passengers in 20157. Another route I know well is Los Angeles-San Francisco. Less passenger volume, but still a significant number of flights. 124 per day in 2013 (Sydney-Melbourne was 192)8.

Sydney-Melbourne and Los Angeles-San Francisco are excellent candidates for automation. They are in areas that have relatively small amounts of cross-over traffic. Carving out an automation air corridor is less complex. They are also along ocean routes. A failure scenario where you need to ditch a lot easier to control and work with automation.

As we start our descent, please return your seat backs and tray tables to their fully upright position

While self-driving cars garner a lot of media, it seems self-flying aircraft may be well ahead. The amazing thing about these scenarios is how close they are to reality. Most of these technologies already exist. It's a matter getting them deployed and approved.

And the significant psychological accepted by the commercial-flying public. In that respect, it has a lot in common with self-driving cars.

In the meantime, it seems unlikely that we'll see mass unemployment of pilots. Mandatory retirement will see 42% of US pilots retire in the next ten years9. If anything, this shortfall will push the automation case forward faster than ever.

Share via the meta page

Filed under AI.

Footnotes