Actually, my opinion on the matter conflicts with most people here. I would say, either give the pilot 100% of decision power, or 0%, and let the computer be solely responsible. I keep thinking of those (few) unfortunate accidents that were caused by clashes between pilots and computers.
Pilotless travel may happen, it may not. I'd like to see it. The Economist happily points out that labour costs are the biggest costs for any airline, that the pilot work rota for long-haulers is very restricted by the need for redundancy (i.e. if a plane is delayed by XX hours, the pilot has to be exchanged - or the flight cancelled, because he would violate working time regulations otherwise) and other costs. Well, that's a nice argument, but not convincing on its own.
Then there's the pilot errors. They do happen. (The Economist gleefully points out that the majority of all accidents are CFIT - controlled flight into terrain) But does that mean a computer would do a better job? Or even cope with highly dynamic situations like emergencies?
It's difficult to assess these things. Advances in artificial intelligence (neural networks etc.) may produce more flexible, clever computer, capable of flying planes even in emergency situations. But there are some things where computers are unlikely to be much help. For example, processing of visual information. When something goes wrong at an engine, for example, the pilot can look outside and at least get a general idea of the situation (for example whether it's still attached, or whether it's burning etc.). A computer would have to rely completely on measurements and instruments and numerical data to make its decision. Which makes it more sensitive to instrument errors. A pilot can choose which instrument to trust, but how does a computer choose? All these things need to be adressed and resolved, and many more. But eventually, a computer with flying skills superior to those of any human pilot is quite feasible, in the long run.
I don't think we're 30 years away from pilotless flight, it's at least 50 years. But eventually, it may happen. And I wouldn't be too opposed, either. After all, we trust computers with driving trains, cruiseships (where the captains are there for the tourists and not for much of the actual driving these days), elevators, spaceships, and UAVs. (Just watched a documentary about the Deep Space ion-driven mission, which was an incredible achievement, IMO)
How would it work? The Economist believes a ground controller would supervise 4-5 aircraft simultaneously (i.e. 1 human pilot for 5 planes instead of 2 for 1 plane). I think air traffic control would slowly be changed from a voice-transmitting system to data, air traffic control would be largely taken over by computers, flying the planes would be done by computers communicating with ATC computers, and humans would be supervisors, both on the ground and in the air. Initially, pilots would still fly along, but without anything to do. Eventually, the lead cabin crew would be transformed to a more empowered role - i.e. able to command the computer to land (in case of air rage, for example) or declare emergency. Perhaps he/she'd even get some rudimentary flight training for taking over in case of total power loss for the computers. (But with a provision that this may only happen if the ground controller sets the system free, as provision against hijackings)
The real challenge is to prove that computers are just as safe, or safer, than human pilots. Once that is a) the case and b) proven, their application to aircraft is only a matter of time. UAVs and spaceships will be valuable tools in improving and proving the technology. Once they log enough missions without errors and prove the idea safe and feasible, the next step will be logical...
The other factor, unmentioned by the Economist, is aircraft shape. Today, planes are very traditional in structure and shape. But look at what aircraft designers are dreaming of: Sonic Cruisers and Flying Wings and other beasts. Neither is a shape that's easily flyable - they'd both have to rely on computers to make flying manageable (much like current combat aircraft are totally unflyable without computers). Now once you have to rely on computers anyway to fly the things, the step to give them total control is only logical. So, my prediction: Pilotless planes will not happen in the near future. But when plane configurations become unconventional (which would require quite some advances in materials before that's likely to happen), their time is near.