What is the primary source of an AI model's knowledge?	The data it was exposed to during its training.
What is the fundamental limitation of an AI model regarding time?	It has a fixed knowledge cutoff and cannot browse the real world in real-time.
What defines the "Capability zone" of an AI model's knowledge?	Topics that appeared frequently, recently, and consistently in the training data.
What types of topics fall into the "Limitation zone"?	Rare, post-cutoff, niche, local, or contested topics.
Name four characteristic knowledge failures of AI models.	Staleness, uneven coverage, inherited bias, and source amnesia.
What three product features are designed to patch AI's knowledge gaps?	Web search, retrieval (RAG), and tool use.
What is the difference between an AI's "Knowledge" and its "Working Memory"?	Knowledge is static (trained data); Working Memory is the current context (prompt/documents).
According to the lesson, what does understanding knowledge unevenness help a user determine?	When to supply context, when to use search, or when to hand off the task.