course / QKV

Model output is a result. It is not automatically knowledge.

Students design checks for generated text, labels, or predictions, including cases the model should miss.

Record the miss, not only the fluent hit

An AI check at QKV is a small evaluation log: where the test items came from, how results were classed, one failure example, a limit, and the next trial. This is verification, not HNK’s literacy explanation. Do not invent a benchmark score for this school.

Failure log

info@qkv.org