DesignationISO/IEC/IEEE DIS 42703
Stage40.99, DIS approved for FDIS
Planning date, an estimate13 Sep 2027
Published source todayIEEE Std 7003-2024
9.1Clause
Evaluation

Purpose

IEEE Std 7003-2024, Clause 9.1Carried into ISO/IEC/IEEE DIS 42703Retrieved 20 October 2026

Clause 9 evaluates the system as a whole for bias through a design and output evaluation and an ongoing evaluation.

What clause 9.1 asks

Two evaluation procedures sit in this clause. One looks at the design and at what the system outputs. The other sets up a routine that keeps watching after the build. They are different jobs. The first belongs to design and development; the second runs through operation and retirement. Each one adds to the bias profile.

Why it matters

The two-procedure structure prevents a common failure. A team runs a pre-release test and treats the evaluation as finished, then the system drifts in production with no plan to notice. Or a team monitors production but never examines the design and the outputs against the requirements. The clause asks for both, and it links each to the profile so the findings are recorded where the rest of the work can use them.

What good looks like

Plan the two procedures together. The design and output evaluation sets the baseline and the metrics; the ongoing evaluation monitors the same questions over time. Record how the findings from one feed the other. Resource both, and name the person who owns each. Then confirm that the evaluation plan references the current profile version.

Evidence an assessor asks for

The assessor will ask how the ongoing evaluation relates to the pre-release evaluation. Two unrelated plans are a sign that the clause was read as two separate tasks.

  • Evaluation plan The design and output procedure and the ongoing procedure, with owners
  • Baseline record The design and output evaluation that sets the baseline
  • Monitoring plan The ongoing procedure and its link to the baseline
  • Resource record The effort allocated to each procedure

Where it commonly fails

  • Only a pre-release test is run
  • Only production monitoring is in place
  • The two plans use different metrics with no explanation

What to do before publication

Design the baseline and the monitoring together, so the same questions are asked at both stages.

Read across

← 8.5 Outcomes9.2.1 Evaluation of design and outputs: purpose →

Full register / The dates / Free scope check