Documentation for AI systems

1. Data

Are there any agreements regarding data delivery?

Low impact

The agreements on data delivery are sufficiently clear. These agreements are well documented.

Medium impact

There are a few agreements on data delivery. These agreements are somewhat documented.

High impact

There are no agreements or documentation regarding data delivery.

checkbox

Checkbox 1

Checkbox2

Text *

Text ARea *

AsfpoASdokas

File upload

Upload

email

2. Personal data

To what extent are personal data used in the algorithm?

Low impact

No personal data are used and/or processed.

Medium impact

Some personal data are used, but the data cannot be traced back to an individual person.

High impact

Various personal data are used with which (directly or indirectly) the person in question can be identified.

3. Consequences for citizens I

Is the algorithm part of a decision-making process that has legal consequences for citizens?

Low impact

No, the algorithm is not part of a decision-making process that has legal consequences for citizens.

High impact

Yes, the algorithm is part of a decision-making process that has legal consequences for citizens.

4. Consequences for citizens II

Does the algorithm affect (or threaten to affect) a fundamental right? If so, how severely is this fundamental right affected?

Low impact

Small to no violation of a fundamental right, so no special degree of dilligence required.

Medium impact

Moderate violation of a fundamental right, so dilligence is required.

High impact

Serious violation of a fundamental right, dilligence in particular is required.

5. Population size

What is the size of the population to which the algorithm is applied?

Low impact

<2.000 individuals

Medium impact

2.000 - 100.000 individuals

High impact

≥100.000 individuals

6. Third parties

Are the outcomes of the algorithm shared with third parties (e.g. citizens, other departments or supervisory parties)?

Low impact

No, the results of the algorithm are not shared with third parties.

High impact

Yes, the results of the algorithm are shared with third parties.

7. Financial/reputational damage

What is the (estimated) financial or reputational damage if the algorithm contains errors that lead to incorrect outcomes?

Low impact

Using the algorithm has minimal financial or reputational impact.

Medium impact

Incorrect outcomes of the algorithm lead to average financial and/or reputational impact damage.

High impact

Incorrect outcomes of the algorithm lead to great financial and/or reputational impact damage. Errors have a big impact on the service of the organization.

8. Autonomy

Are the decisions made based on the outcomes of the AI system? If so, how?

Low impact

Output of the AI system is descriptive or diagnostic statistics, which serves as input for decision making.

Medium impact

Output of the AI system is predictive statistics, which serves as input for decision making.

High impact

Output of the AI system is prescriptive statistics, or decision-making is fully automated.

9. Validation

Are the results of the algorithm validated?

Low impact

Yes, the AI system has been validated.

High impact

No, the AI system has not been validated.