"The mate's AI uses a few simple actions: approaching, circling, attacking, and sprinting away. It tries to get behind the elk, though in this one-on-one case it's pretty hard. Every now and then it'll try to attack, but might be scared off mid-way by the elk's counter-attack (with or without getting kicked).
The tricky part is building the AI calculations ("scorers" in Utility AI terms) around when exactly to do each of these things. How often should the mate try to attack, and how much risk should it take? What should it take to scare it away? Should it always keep circling in one direction or frequently change the direction? How far should it circle? If it does land a bite, how long should it keep it?
And how does personality (mainly "Cautious - Bold" but also the other axes) influence this? We decided to give both ends of the "Cautious - Bold" spectrum their own pros and cons. A Bold wolf will waste less time fiddling around while a Cautious one will be smarter about timing and target selection. We also want to use uniform mechanics for other fights, so we can use the same / similar AI."