Quality before trust

1. 01

   Acceptance criteria

   Clear conditions a deliverable has to meet before anyone calls it ready.

2. 02

   Adversarial review

   Look for contradictions, gaps and the situations where the result could fail.

3. 03

   QA: quality assurance

   Organise checks to reduce defects before releasing work.

4. 04

   CI: checks with every change

   Run checks automatically every time a project changes, so problems show up early.

5. 05

   Evaluation and reliability

   Test results against set cases and criteria, and keep watching whether quality holds over time.

A convincing deliverable still needs checking. You learn to look for failures, test what matters and keep a record of the evidence.

• Define observable criteria for approving a deliverable.
• Apply a critical review and record the issues found.
• Choose checks proportionate to the impact of the task.

1. Set three approval conditions before execution.
2. Examine the deliverable and record the failures found.
3. Correct it, check again and preserve the review evidence.

Fill in the blanks with your own work. This is the deliverable you take out of the module.

Acceptance checklist; critical review guide; evaluation and corrections worksheet.

Planned for the classes of this module.

L04 and L05 for the complete exercises; criteria and review also apply to L02 tasks.

Choose this ifI want to assess quality and reduce failures in AI deliverables.