Produce random copies of integer list

my project at university called "Optimal Quiz Maker"The teacher enters an unlimited number of questions of a specific course, and append to each question unlimited number of answers (multi-choice).The system should produce exams "diffrent copy of question distribution" that must be unique as possible, where each student should take a diffrent distribution of questions and answers, *"we know nothing about the exam paper of the studen sitting next to him.*I designed an algorithm that guarantee a perfect random distribution> but still don't have a reliable metric, that evaluate distribution of all exams.

Example:

exam1: [1,44,56,3,7,13]

exam2: [2,44,15,24,13,7]

Note that we satisfied a good randomness in level of one exam, but we have two similar questions

Note that the complexity increase when we consider the answers in distribution.> what is the perfect distance metric that perform max entropy for all copies of exam?

what method or idea can give the max entropy of all copies of the exam?

Is it useful to use stimulating annealing or genetic algorithms

Sorry for every unclear point, I'm grateful for any help.