During the roundtable discussion, the participants concluded that establishing an AI ethical board can help build customer trust, ensure responsible and ethical use of AI, and assist with navigating increasing regulations. Organisations, regardless of their size, can benefit from some form of ethical oversight, whether through a dedicated board or ad hoc advisory. AI ethical boards can provide independent insight of AI initiatives, and advise on ethical considerations throughout the lifecycle of an AI project.

However, the need for an AI ethical board, and what it should look like, will depend a lot on the organisation’s specific context. For example, it might be more relevant in general for governmental and B2C entities.

What to consider when setting up an AI ethical board

The participants brought up a number of possibilities to consider when setting up an AI ethical board, such as:

  • ensuring independence, whether the AI ethical board is set up as a separate entity or as part of another entity
  • choosing to have the AI ethical board report to the Board of Directors
  • including both internal and external members, from diverse backgrounds and areas of expertise, to provide different perspectives and advice
  • implementing consensus-based decision-making, so that input from all members is taken into account
  • organising regular meetings, with a frequency based on your organisation’s specific situation and needs
  • drafting a document outlining governance processes, decision-making processes and procedures
  • ensuring the need for resources such as access to relevant expertise, training and education, administrative support, and support for the board’s activities and initiatives.

Real-life case: VDAB

Our host, VDAB, has set up an Ethics Council that offers an excellent example of the importance of independent oversight, risk classification and regular review of the guidelines:

  • VDAB uses a 4-phase roadmap for AI projects: Experiment, Explore, Exploit and Execute
  • The Ethics Council oversees both experiments and projects, providing input starting at the ‘Experiment’ phase
  • Due to time constraints, only projects with the highest impact/risk are presented to the board
  • The Steering Committee meets monthly, and decides which ideas will move forward to the Experiment phase. However, small-scale experiments are always permitted, to encourage innovation
  • More than 50% of members are external, to ensure impartiality and offer diverse perspectives
  • The board meets every 3 months

Find out more from the VDAB presentation from our “Ethical AI culture: governance & acceptable use in business” event in February 2024 (available for Beltug members after logging in).