The concept of sensitivity is actually not too far from the questions asked in ML. "The minimum number of bit flips needed to change the answer" -> adversarial

The concept of sensitivity is actually not too far from the questions asked in ML. "The minimum number of bit flips needed to change the answer" -> adversarial attacks, prediction stability, corner case performance. "Which bit flips are needed to change the answer" -> feature importance.