Focusing on foundational system behavior rather than specific agent/model frameworks. Teaching how inputs introduce bias and shape non-deterministic outputs is relevant across model generations.
This is relevant for how you craft and input and how the LLM provider does on their end