The classification of classifiers

Posted here (week 2442).

Amsterdam has a public registry of algorithms that is a very promising example of civic transparency and algorithm/AI accountability. Recorded here for posterity are the considerations listed in this catalog today (late 2020) and their descriptions:

Datasets

Key data sources utilised in the development and use of the system, their content and utilisation methods.

Data processing

The operational logic of the automatic data processing and reasoning performed by the system and the models used.

This includes architecture overview, source code, and performance measurement.

Non-discrimination

Promotion and realisation of equality in the use of the service.

Human oversight

Human oversight during the use of the service.

Risks

Risks related to the system and its use and their management methods.