Unless there is some fundamental change in the underlying technology (and I’m more than happy to stand corrected on this), the sophistication and complexity of hardware required to support a true artificial intelligence will be staggeringly expensive and require continued (human) maintenance. Unless you can develop smart system ‘drones’ to undertake repairs, or have two AI per ship, so that one can act as doctor for the other if required. That’s a lot more money, squared.
Human life support is a given, a known and quantifiable variable, with predictable costs over the long term. Plus I believe that the human ego will dictate having the final say, regardless of the advantages an autonomous agent would bring (in reaction time, ethical dilemma, indifference to casualties, etc.)