Which of the following situations might make reinforcement learning a more suitable approach than supervised learning? (Mark all that apply) Group of answer choices When the skill of the AI must exceed the human "experts" available to train it. When the environment is dynamic and constantly changing. When the problem involves labeled data that is easily available and straightforward to use for training. When the optimal behavior depends on a sequence of actions and not just a single input-output mapping. When rewards are not easily defined When there are many possible states and actions, making it difficult to design explicit rules or algorithms