When I started my military service in the Israeli Defense Forces, I was a candidate for pilot training. I was on a path to fly F15s, C130s, or Helicopters. In Israel, being a pilot was considered one of the most prestigious occupations. The main Israeli airline, El-Al, had no more than 20 planes at the time. The Israeli Air Force trained many more than 20 pilots every year. As a result, you had to know someone to become an airline pilot in Israel. Because of that, I opted out of pilot training.
In the US, the situation is different. Boeing's vice president of product development, Mike Sinnett, said in an interview that he expected a need for 41,000 commercial airliners over the next 20 years, and all of those would require pilots. Pilot training is expensive. Getting a private pilot's license costs almost $10,000, but the cost of training to become an airline pilot could exceed $50,000. With the median pilot pay of $102,000, fewer people are taking this route.
One way to solve the problem, Sinnet said at the Paris Air Show last week, is pilot-less planes.
Planes today fly using auto-pilot for majority of the duration of the flight. GPS and radio navigation (VOR) are used in conjunction with the auto-pilot to fly the plane accurately from one waypoint to another. Most commercial airliners are also equipped with auto-landing systems that can autonomously land the plane, typically in very low visibility conditions. What's to prevent planes from being completely autonomous, then?
The psychological factor.
Would you trust a machine to fly you from place to place? There needs to be a leap of faith by passengers in order for them to trust their lives with a computer. After all, if there is nobody at the cockpit, who will press Ctrl+Alt+Del if the flight computer stalls? (no pun intended...) There are two answers to this concern.
First, there doesn't have to be a direct jump from two (or three) pilots to none. The number of pilots can be reduced to one, initially. The artificial intelligence system could do a lot to reduce the load off the pilot.
Second, the same could be said for autonomous vehicles. A decade ago, nobody in their right mind would consider traveling in autonomous cars. Today, they are. Once the visionaries and early adopters show the less-adventurous users that self-driving cars are safe, reliable, and efficient, adoption accelerates. The same could thus apply to commercial aviation.
Initially, autonomous jets will likely be limited to cargo planes. Packages, after all, are much less likely to object to being flown by robots than humans are. As more and more cargo planes fly autonomously, the more passengers would be willing to consider flying in autonomous jets themselves.
The regulatory factor.
The FAA has, for a few years now, been investigating autonomous aircraft. The main concern is that an autonomous aircraft would not be able to observe conflicting traffic and might cause a fatal air collision. In the FAA's study, the main technological barrier is the aircraft's ability to "detect and avoid" other traffic. Not until it is proven that autonomous planes have a "detect and avoid" capability equal to, or better than human pilots, that the FAA would allow flying autonomous aircraft in the national airspace. For that reason, when the FAA released new rules that allow the use of drones commercially, they limited them to line-of-sight operation, which would not be practical for long-haul commercial aviation. However, those rules apply only to visual-flight-rules (VFR) flying. Commercial aviation typically flies in Class-A airspace (above 18,000 feet), where VFR does not apply. Planes can fly above 18,000 feet only if they are continuously monitored and controlled by Air Traffic Control (ATC), and carry radars and identification systems (transponders). The main responsibility of ATC is to maintain separation between planes. Pilots are the second line of defense. Under those conditions, autonomous aircraft become more feasible, as long as they are constantly monitored and controlled by ATC, although regulations (and ATC) will have to change to support such operation. Much like allowing drone operations and autonomous car testing, FAA waivers would likely be required initially, until those rules change.
Either way, Boeing plans on testing autonomous commercial airplanes next year.