Action selection in AI systems is a fundamental process where the AI analyzes the problem to determine its next step towards solving it. We will be working with C++ at AIPuzzles due to its performance advantages. While AI is often developed and taught in Python for its ease of use in teaching, learning, reading, and writing, C++ has significantly evolved, especially in its syntax. We will be programming in C++17, which offers syntax improvements comparable to, or even surpassing, Python. Pay close attention to the details.

The agent, in this context, refers to the programmed artificial intelligence that enforces a policy (state -> action). It is responsible for generating an appropriate action based on the state received from the environment. In more complex scenarios, like reinforcement learning, the environment also provides feedback in the form of rewards based on the agent's previous actions.