Unfortunately you seem to be conflating a two distinct issues in one question.

1. Computer control of aircraft.
2. Unstable aircraft design.

First dealing with computer control, or flight control systems. Many modern designs military and civil are using digital fly by wire (FBW) systems. This has little or nothing to do with aircraft stability. These systems often have no mechanical backup, so in reality the pilot is merely instructing the computer how to fly the plane. Modern FBW systems use a quadraplex system, i.e. there are 4 duplicate computers. 3 are active and one is hibernating. Then if one system goes duff the other 2 can switch the duff one off and bring the 4th out of hibernation. If you only had 2 computers how do you know which one is wrong? - hence 3 are used. Prior to flight clearance these systems are tested with thousands of hours of rig testing, safety checks etc. Whole volumes have been written on the subject.

Second issue - unstable aircraft design. Again a couple of comments, first aircraft instability has nothing to do with manoeuvrability - modern military aircraft are aerodynamically unstable for supersonic performance not manoeuvrability. Modern FBW systems allow designers to explore unstable designs but there is a limit. Originally the FBW systems were looking at pitch instability but now it is being extended to lateral instability for some of the more novel stealth designs.

So in terms of your two proposed conditions - if there is a mechanical backup then pilots can adapt and are capable of controlling slightly unstable configurations in the event of a failure, although they may whinge a little about the handling qualities. For a fully digital FBW system then for your proposed conditions then as the systems failed there would be a gradual degradation in flying qualities but there would still be a backup system allowing a return to base capability. To have a total loss in control then there would have to occurred a series of multiple failures. Again risk and safety assessments are another whole volume of work.

Unfortunately you seem to be conflating a two distinct issues in one question.

1. Computer control of aircraft.
2. Unstable aircraft design.

First dealing with computer control, or flight control systems. Many modern designs military and civil are using digital fly by wire (FBW) systems. This has little or nothing to do with aircraft stability. These systems often have no mechanical backup, so in reality the pilot is merely instructing the computer how to fly the plane. Modern FBW systems use a quadraplex system, i.e. there are 4 duplicate computers. 3 are active and one is hibernating. Then if one system goes duff the other 2 can switch the duff one off and bring the 4th out of hibernation. If you only had 2 computers how do you know which one is wrong? - hence 3 are used. Prior to flight clearance these systems are tested with thousands of hours of rig testing, safety checks etc. Whole volumes have been written on the subject.

Second issue - unstable aircraft design. Again a couple of comments, first aircraft instability levels are not driven by manoeuvrability requirements - modern military aircraft are aerodynamically unstable for supersonic performance not manoeuvrability. Modern FBW systems allow designers to explore unstable designs but there is a limit. Originally the FBW systems were looking at pitch instability but now it is being extended to lateral instability for some of the more novel stealth designs.

So in terms of your two proposed conditions - if there is a mechanical backup then pilots can adapt and are capable of controlling slightly unstable configurations in the event of a failure, although they may whinge a little about the handling qualities. For a fully digital FBW system then for your proposed conditions then as the systems failed there would be a gradual degradation in flying qualities but there would still be a backup system allowing a return to base capability. To have a total loss in control then there would have to occurred a series of multiple failures. Again risk and safety assessments are another whole volume of work.