For goodness sake, Airbus have been banging on about this nonsense since the 1980's.
The stupid detent-type thrust levers (why cant those things just move normally) were originally designed as power switches ffs.
Airbus have succeeded in removing pilots from the loop but not eliminating them and in doing so created a messy no-man's land in the process.
But could computers really do it? How could a computer perform ECAM, QRH, OEB, FCOM, MEL or whatever sequence it is this week? How could it understand Fringlish? How could it ever work out those ridiculous landing distance charts? Does a computer understand an asterisk?
Yes there are many pilot-induced accidents but for at least the next 80-100 years we will never solely trust computers. One major issue is that pilots have to try to second-guess what the bloody thing is doing when failures overwhelm limited computer capabilities. How would the Qantas engine explosion have panned-out if the pilots hadn't have been there? Reckon the best the MCDU would have come up with is CTRL+ALT+DEL
Tried doing an autoland in an A330 in what the computer regards as confusing winds (a little downwind or crosswind)? Unlike the 747, its awful. Had to take over on a number of occasions as the thing pulled off trust too high and hurtled earthwards.
Prediction: Year 2085: "European/Chinese Aircraft manufacturer Airbus SAS says that it is close to perfecting a pilotless ducted-fan A390. The aircraft has only been approved to carry animals and freight and can only fly over unpopulated areas and land at airports at least 100km from the nearest city. EasyRyanair have expressed an interest".
Good luck with these plans Mr Airbus - laughed 30 years ago when first heard the story and nothing has swayed my opinion about the stupidity of this concept.
Now off to feed the dog that's there to bight my hand etc etc.