How can we design assignments and exams that meet our learning objectives and discourage inappropriate use of AI?