Hello everybody! I am thinking of building a generic reinforcement learning platform as a service and I think it has applications in games. Specifically either building AIs to play along/against humans and adapt difficulty/skill of AI on the fly, or model a way to adapt difficulty based on the skill of the player. How I think it would work is as follows: the game sends a list of potential actions and a game state (for each AI agent) to the platform (can run locally), and one action is picked. The AI agent performs that action, and sends back an error measure (e.g. new game score, health level, etc.) and the cycle repeats. Models can be pre-trained so they don't play too dumb the first time. The project is at the very beginning. I would like to ask you, the indie game developers, if this is something that you would find useful at all? Appreciate the time for reading this and waiting for your feedback or suggestions! Thanks!