Trustworthy AI Project Planner

Developing an AI system involves a lot of work: technical choices, data infrastructure, legal aspects... But how do you ensure that your project is ethically sound from the outset, and how do you prioritise what to focus on first?

This short list of questions helps you consider the most important ethical considerations at the start of an AI project, based on the European Trustworthy AI framework. 

Why use this planner?

The planner is a practical tool for structuring ethical reflection from the outset of an AI project. By consciously considering eight ethical principles, you, as a project manager or product owner, can quickly assess:

• Which principles are relevant to your project
• When to start working with them
• What actions are needed
• Who is responsible for follow-up

The result is a clear action plan tailored to your AI project, including a prioritisation of the most important points for attention.

There is a longer self-assessment list that you can use, but it consists of more than 100 questions. This planner is shorter and helps you identify the priorities according to the Trustworthy AI framework.

When should you use this planner?

The planner is intended to incorporate ethical considerations into the design, development and planning at an early stage. Think of it as a guide to project design.

You can integrate this spreadsheet into the ethics section of your project proposal or project initiation document. This allows reviewers to see at a glance how you are applying the relevant principles.

Who is this planner intended for?

For project managers, product owners or similar profiles who are responsible for starting up and coordinating an AI project. We recommend completing the questionnaire together with a multidisciplinary team, for example with colleagues who have technical, legal, social or ethical expertise.

The planner is based on the seven principles from the ALTAI framework of the European High-Level Expert Group on AI, with additional attention to fundamental human rights in an eighth principle:

1. Human agency and oversight
2. Technical robustness and safety
3. Privacy and data governance
4. Transparency
5. Diversity, non-discrimination and fairness
6. Social and environmental well-being
7. Liability
8. Fundamental rights
    

For each principle, the questionnaire contains two reflection questions. These questions refer to the current iteration you have in mind for your project. For example, when creating a proof of concept that will not be put into production. This means that certain ethical requirements are not yet relevant because they can or must be considered in a later iteration. During a proof of concept, robustness and cyber security may not yet be included because this only becomes truly relevant during implementation. In other words, the idea is to make a plan and prioritise what can and must be done in a particular iteration.

The planner is an Excel file that you can fill in together with your team as you go along.

1. Read the two questions for each principle; these will help you to think about the possible ethical impact of your AI system.
2. Fill in the columns for each question (see below).
3. Discuss the answers with your team to arrive at a shared understanding and plan of action.
4. Save the document so you can reuse it later as a reference, for accountability or for evaluations.

What do you fill in for each column?

• Relevance: Does this principle apply to your project? Choose, for example: “High”, “Medium”, “Low”, “Not relevant”.
• Timing: Indicate when you will address this aspect. Choose from: “already implemented”, “will implement soon”, “will implement later”, “do not plan to implement”.
• Actions: Note down the specific steps you will take to apply this principle.
• Responsible person & deadline: Indicate who within or outside the team is responsible for following up on these actions.
• Comments: Use this column for additional explanations: why something is (not) relevant, what choices have been made, or where there are still uncertainties

This planner is not a checklist, but a tool to make ethics workable from the very beginning of your AI project. Use it as a stimulus for discussion, collaboration and critical thinking.

This is a planning tool for prioritisation, so it is important to be able to determine what is relevant and can be resolved now, but also what can be addressed later, so that the list of ethical requirements becomes achievable and realistic.

This planner is available under a CC BY-NC-ND 4.0 licence. This means that this tool may be shared under the following conditions:

• Attribution: You must correctly attribute the licensor, Kenniscentrum Data & Maatschappij, include a link to the licence and indicate whether any changes have been made. This may be done in any reasonable manner, as long as it does not give the impression that the licensor endorses you or your use.
• Non-commercial: You may not use the material for commercial purposes.
• No derivative works: If you edit, adapt or build upon the material, you may not distribute the modified material.

Image by Hanna Barakat  & Archival Images of AI + AIxDESIGN via https://betterimagesofai.org - license CC BY 4.0