I have read with some surprise that the US Navy has set the objective that 60% of the air wings of its future aircraft carriers were drones. It is true that the journalist who wrote it (quite reputable by the way) did not advance a date for this event, but it gives us an idea of what military strategists are thinking for the near future: the massive use of drones and AI.
The USA have the most experience in the use of military drones, and have the most advanced drones, so they have real experience of what a military aerial drone can do compared to a piloted military aircraft. They are devices that cost much less and do not put the lives of their pilots at risk, which require many years of learning.
Apparently, strategists predict having 60% drones with autonomous systems on a carrier, and 40% manned aircraft. And that could happen in the next 20 or 30 years.
There are already various military drones such as the Walkyrie with certain autonomous capabilities and that can act together with other manned aircraft such as the F-35C. If we think that an F-35C costs about $100 million, and the Walkyrie less than $10, we see that the combination of both (or similar ones) is more than interesting in terms of costs.
But in matters of Defense, the costs are very important, but they are not the most important thing. The most important thing is effectiveness, and today drones need to be controlled by people. It is therefore essential that these drones acquire capabilities to operate autonomously, and they will do this by incorporating AI into their control systems, and making those systems fail-safe.
Drones equipped with AI to act autonomously and in coordination with other aircraft introduce a certain "uncertainty", which must be resolved in the coming years based on real tests in different operational circumstances.
Today, operations with pilots have a high degree of efficiency, and that is the level that drones with AI must achieve. At least the objective is clear and they have a way to measure it, which is an important help when making decisions by those responsible.
The US Navy has already tested some drones flying from an aircraft carrier before 2015, and they are currently testing new models, I suppose to gain new experiences.
Adding AI to a military drone operating autonomously from an aircraft carrier is very, very complex. Starting with the ability to shoot down enemies (it is a machine that makes the decision to kill a person), going through the cooperation of actions with other planes, to be able to take off and land on an aircraft carrier is very complex, etc…. and all this without computer or other failures.
We are still a bit far from that situation, but not too far; and that's why military strategists are thinking about it. It's a question of time.
EN 
ES_ES 