Evaluate, Revise and Disclose
What this concept is
A small evaluation is a set of predeclared checks applied to a response. Disclosure records which work was yours, which came from a tool or fixture and how you checked it. Both make a polished result inspectable.
Why it matters here
The aim is to become a better learner and judge of work, not to report how many prompts you sent. A response should help achieve a defined task while preserving the skill you are practicing.
Concrete example
Before reading a plan, write four criteria: the minutes fit, the supplied facts are preserved, unsupported claims are flagged, and no unauthorized action is claimed. Candidate A may satisfy three and fail one. Do not call it “75% safe”; those four checks are not a calibrated risk model, and a serious failed condition may require rejecting the plan.
Common confusion / misconception
One response does not establish a model's general quality. Authored offline candidates are not sampled model outputs. Editing a response after seeing the answer key makes it a repaired result, not an untouched first attempt.
How to use it
Save the original task, your attempt, assistance type, the candidate, check results and revision. Add an independent changed-case attempt. Record product/model/date only if actually observed during a live session. For an offline exercise write “authored fixture; no model called.”
Check yourself
Can a learner pass without an AI account? Yes: evaluate the supplied candidates, design a bounded tutoring request, repair a failed plan and independently solve the transfer task. That demonstrates workflow judgment, not live product fluency.
Mini drill or application
Write a five-line assistance declaration and a one-paragraph verdict on the best candidate. State one remaining limitation and one condition under which you would choose to work without AI.
Read this only if stuck
Use the module rubric for observable criteria. Later AI Literacy and Assisted Engineering adds executable evaluators after programming prerequisites.
For broader responsibility and evaluation guidance, consult the diligence/discernment sections of AI Fluency.