Machine Learning to test optimal Raid configurations/Boss behavior

Kybian

After seeing the Dev Question related to NPC behavior, and listening to an interview with Jeffery Bard, I was curious if anyone knew what systems Ashes of Creation were implementing for their dungeon/raid boss behavior. Jeffery stated that based on the clear speed of a group, later sections and bosses would pull from a higher/lower moveset difficulty table to scale fights and keep them fresh. As an aspiring Machine Learning Engineer for MMO's, I'm curious in what way they go about handling those complexities. I also wonder if you can take those tables, and run AI against them, to predict the distribution or optimal makeup of the parties for each difficulty table. That way you could test all 64 subclasses without having to wait for beta launches. Thoughts?

Wandering Mist

I very much doubt they will give us that information for fear of people gaming the system exactly the way you describe.