Brilliant Failures Award AI Coalition Public Services

Artificial Intelligence (TO THE) offers many opportunities for optimizing public services. Taking advantage of these opportunities often involves trial and error. When innovating in AI in the complex government domain, it is often (almost-)failures that drive progress. A Brilliant Failure is a well-prepared attempt to achieve something with a different outcome than planned. Failures are brilliant when they are learned from and the experiences are shared with others. We find the opportunities to learn when things go wrong or (upon success) could have gone wrong, but where that didn't happen. You can because you were lucky, but also because you consciously made the right decision, based on thinking, knowledge use, to collaborate, etc.

Presentation of Brilliant Failures Awards AI in the public sector 2023


On Wednesday 12 april 2023 the Awards for the most Brilliant Failure in the field of AI in the public sector were presented for the second time during the Dutch AI congress. This election is an initiative of the Dutch AI Coalition in collaboration with the Institute for Brilliant Failures.

Creating awareness for the importance of learning and experimenting with AI is the reason to pay attention to AI projects in a positive way. The possibilities with AI are enormous, but it does not happen automatically and it is important to learn from all experiences, so that we can use socially relevant applications of AI efficiently, effective and responsible manner. Two prizes have been awarded, an audience- and a jury prize. The jury prize was awarded by Paul Iske, jury chairman, and also Chief Failure Officer of the Institute of Brilliant Failures. The public award was presented by Marieke van Putten, chairman of the Public Services working group at the Dutch AI Coalition.

Jury Prize

The jury prize was won by the Foundation My Data, our health, that works on trust in research data generated and managed by citizens. Core function is connecting requests to (her)use of health data with the person to whom that data relates, simply organizing permission. But even though the data is available and citizens are happy to share it for research purposes, In practice, this is not always easy. In a project, the gap with the Medical Ethics Committee on views and concerns about ethics became apparent, methodology and health insurmountable, and the project had to be stopped. Wet- and regulations are not always in line with developments and wishes from society.

Audience Award

The audience award was won by the Provincial Data Lab North Holland, that wanted to get a grip on tourism flows in nature with the winning project- and recreational areas and learn how the province can influence them based on data and AI. However, the data turned out to be not entirely valid. For example, the recreational area of ​​the Amsterdam forest turned out to be very popular, especially in the morning and afternoon during rush hours. It therefore turned out that the A9 highway, which runs right through the Amsterdam forest, included in the counts. The conclusion was that it is important to involve people who are familiar with the situation.

The importance of a learning approach

The Institute for Brilliant Failures and the Dutch AI Coalition would like to thank all parties who submitted a case for being open and sharing their experience to learn from this approach. We congratulate the Provincial Datalab North Holland and Stichting mijn data, our health with the award of the Brilliant Failures Awards! A more detailed description of the winning cases can be found here here, these will soon also be available on the websites of the Dutch AI Coalition and the Institute of Brilliant Failures can be consulted.

Awards in 2024

The Dutch AI Coalition and the Institute for Brilliant Failures underline the importance of experimenting with AI and express that (brilliant) failure in the responsible development and application of AI is inevitable. We therefore invite all parties in the public sector to compete for the Brilliant Failures Award for 2024.

Jury assessment criteria

The assessment criteriaThe jury selects winners based on the VIRAL formula and considers the following matters:

  • Vision: The extent to which the failed project is based on a detailed vision of AI implementation
  • Effort: The extent to which one has committed to the success of the project and others
  • Risk management: The extent to which one has succeeded in striking the right balance between avoiding unacceptable risks, but daring to take acceptable risks
  • Approach: The extent to which one has prepared sight, cooperated, has used available knowledge
  • To learn: The extent to which people have learned from this project themselves and the knowledge has been or can be shared with others

Procedural criteria

  • Submission is through 31 March 2023 sent via the registration form on the right for the 'competition brilliant failure AI in public services
  • Transparent and compact information about your own position and specific role in the process
  • Email address and telephone number of sender known
  • The entrant has checked with his/her organization whether the case description may be made public in the run-up to the assessment of the entries, and announcing the winners.
  • Submitter's motivation why this failure matters, why does it have so much mhonor to share (e.g. as a closing word)
  • The project was carried out by party(and) in the public sector. Those are ministries, implementing organizations, municipalities, provinces, regional water authorities, collaborative organizations and not for profit organizations that perform public tasks.

Composition of the jury

  • Prof. dr. Paul Louis Iske, Chief Failure Officer, Institute for Brilliant Failures, chairman of the jury
  • Drs. Robert van Doesburg, Senior Scientist, TNO-Leibniz Institute, TNO
  • Ir. Marieke van der Putten,
    Senior Innovation Manager NL AI Coalition
Award uitreiking zorg 2020