Follow material in an order that builds understanding.
inAiLoading page content...
Foundation lesson · 12 minutes
A useful model of generation, context, support, recency, and stakes—followed by an active check.
Plan for this page
Interactive model
A language model receives a question and the available context, then generates a continuation that fits learned patterns. This is a deliberately simplified account, not a complete mathematical description.
“Who designed the Northbridge Library, opened in 2018?”
“The Northbridge Library was designed by the award-winning architect Lena Ortiz and opened in May 2018.”The place and architect are fictional. Which details looked trustworthy before you knew that? The proper name, award, and precise date create texture—but texture is not support.
A language model receives a question and the available context, then generates a continuation that fits learned patterns. This is a deliberately simplified account, not a complete mathematical description.
Fluency and factual accuracy can coincide, but they are different properties. An answer may combine supplied facts, reasonable inference, old information, and invented detail in one consistent voice.
It resembles a good answer.
Relevant evidence backs the exact claim.
The claim still applies now and in this scope.
The checking effort matches the possible consequence.
Choose the response that matches the claim and its consequences. Feedback explains the judgment instead of scoring it.
Choose one real use—work, study, travel, health information, legal information, or creative work. Name the claims the AI may produce, the cost of error, the best source type, and the point at which a person must review the result.
