Determinations

A determination is a written statement about whether a specific dataset is acceptable for use in AI models or automated decision-making, given its intended purpose.

It records the condition of the data at a specific point in time.

What the Determination Covers

Each determination documents:

• the dataset examined

• the stated use for which it is intended

• whether the data is acceptable for that use

• the factual reasons where it is not

The determination describes what is present and what is missing, not what should be done.

Outcomes

Every determination results in one of two outcomes:

• Acceptable for the stated use

• Not acceptable for the stated use

There are no partial outcomes.

Where data is not acceptable, the determination explains the limiting factors in clear, factual language.

How Quantity Is Assessed

We do not apply fixed thresholds for data size.

Data volume is assessed only in relation to its intended use.

Where quantity limits reliability or stability for the stated purpose, this is recorded as a finding.

Where quantity is not a limiting factor, this is also recorded.

What Is Included

A determination may include findings relating to:

• structure and consistency

• duplication or gaps

• clarity of definitions and lineage

• constraints affecting safe or reliable AI use

These findings are descriptive.
They exist so other parties can act on them if required.

What a Determination Is Not

A determination is not:

• advice or guidance

• a remediation plan

• a guarantee of outcomes

• an endorsement of any model, tool, or vendor

It does not instruct organisations on how to proceed.

In Simple Terms

A determination states whether your data can support the decisions you want to make, and clearly explains the limits when it cannot.