This always encourages a human strategy where the optimal behavior is to sandbag in whatever way the AI is measuring your progress, and then slingshot past them once the opportunity arises. Not to say that can't be fun for some people, but it's definitely not a silver bullet.