Skip to content

Introduction to AI

Estimated time to read: 4 minutes

Note

Please refer to this repository in order follow the previous assignments for the first course of AI. https://github.com/InfiniBrains/mobagen

Topics suggested in the survey, and some of my considerations.

  • Procedural Content Generation. Advanced terrain generation - It was previously covered in the last class, I am going to focus other topics
  • AI applied to improve 3D Animation Movement. Follow this https://github.com/sebastianstarke/AI4Animation
  • Topics relating to an AI Fighting game - Mostly Agents, State Machines and latency simulation (reflex)
  • Tactical AI - Linear programming, Restriction and Satisfiability problem
  • Neuron networks / Machine learning - This can be real hard to cover all topics in this class
  • Genetic algorithms and Reinforced learning - Find the best parameters for agent behaviors
  • Chess AI - In a broader sense it is a table game, and it is mostly heuristics and state exploration, chess is awesome to learn optimization techniques to reduce memory usage, space exploration, branch and cut, minmax, planning and satisfaction
  • Prediction algorithms for multiplayer - We can cover some techniques to extrapolate data to compensate lag instead of just mathematically extrapolate position, this is mostly an application of agent theory.
  • Stable diffusion/chatbot - This is a hot topic, I didn't went too deep on that, but I can help you at least surf this wave to create fun stuff for games, such as dialog creation.
  • Procedural audio generation - Most of them use convolutional networks mixed with recurrent neuron network. It can be real hard, so if we cover that, we are just goint to understand the overall idea, and learn how to use pre-determined models available for free.
  • Behavior trees - I have to be honest this is a topic that I don't like, but it is a good tool to have in your toolbox, so I can cover it.
  • ChatGPT and its siblings to generate text - I can cover at least how to modify small scoped model and use for your own intent.
  • Stable Diffusion and its siblings to generate images - I can cover at least how to modify small scoped model and use for your own intent.
  • AI subsystems and how to debug it.
  • Spatial quantization optimized for AI queries - I really enjoy this, but it can be hard to understand, because it uses lots of data structures

Note for myself: game worldbox