There are many different ways of making an AI respond to the conditions it is given.
In the assessments of Crypt of the NecroDancer and Duel Links, there are two different primary methods of decision making. Duel Links more than likely implements decision tree algorithms to decide on the best card and how to play it on that turn whilst NecroDancer defines a predetermined set of constraints (such as movement direction) and makes enemies check the appropriate tiles to determine a valid move, which is then all done to the time of the beat. This is essentially a constrained version of path finding, akin to using A*.
Using the appropriate method, or methods, in creating AI for this project is essential and as such, needs to be designed effectively in order to make believable and efficient AI enemies.
Decision Trees
http://blog.hackerearth.com/game-playing-programs
http://www.solipsys.co.uk/new/DecisionTreesInGames.html?InternalLinks
https://en.wikipedia.org/wiki/Decision_tree
https://www.usna.edu/Users/cs/roche/courses/s13si486d/u03/
https://en.wikipedia.org/wiki/Finite-state_machine
https://gamedevelopment.tutsplus.com/tutorials/finite-state-machines-theory-and-implementation–gamedev-11867
Aaron Swiss-Hamlett 