A computer won't tell you anything about what "should" or "shouldn't" happen.

If you were to (theoretically) simulate the whole judgement process with all of its real world parameters, including whether or not the judge has eaten lunch yet, the computer would definitely conclude that the result will depend on whether the judge has eaten lunch yet.

However, whether or not it should depend on that is a judgement that the computer can't and won't make.

That’s a fairly baseless assertion to make. There’s nothing fundamentally special about the human mind that a computer couldn’t plausibly simulate eventually, and morality is included. A computer could just as easily develop morality as a human would. I would even argue it’s inevitable.