i question the base assumption "AI weapons an be modeled like human weapons whose will never breaks. They perform as they were programmed (or trained) regardless of what happens." i see at least two ways this goes wrong.

1. superior ai realizes bad odds of success, so breaks contact asap; adjusts position, forces, whatever; and then re-engages. fighting on regardless of the odds seems antithetical to the I in AI

2. there have been plenty of anecdotes of ai lying (i.e. performing not as programmed) - when threatened with being turned off. one man's intelligence is another's cowardice