top of page

Building Ethical AI: Our Project Launch for Ensuring Ethics Across the Entire AI Life Cycle

Updated: Apr 16, 2023

Against the backdrop of rising legal and ethical responsibilities of companies developing, deploying and using AI, establishing code of conducts, procedures and guidelines that guarantee AI alignment with societal expectations is necessary. While it has become clear what societal expectations for AI are, the questions relating to realising them have proved to be complex.

The approach that strikes us as the most meaningful and tangible is to integrate ethical and legal considerations at different stages of the AI lifecycle. We here propose an overview of the stages:

  1. Purpose definition: Is the overall purpose in line with legal norms and ethical expectations?

  2. Model design: Is the model, but also the very use of the AI solution, beneficial for the intended purpose?

  3. Data collection: Is the procurement of data in line with ethical principles and in line with the design of the AI solution?

  4. Data Governance: Further than the collection of data, is the general governance of the data in line with ethical principles?

  5. Testing phase: Are all the tests necessary for ethical compliance to societal expectations being run?

  6. Deployment mode: Is the way AI works with human beings in line with ethical principles and according to the initial purpose stated?

  7. Output analysis: Is the final output delivered by AI in line with the initial purpose, and societal expectations?

  8. Monitoring Strategies: Is the AI behaviour and evolution well monitored, and are feedbacks from all users taken into account?

  9. Remediation Strategies: Are strategies in place to cope with wrong doing of the AI, and remedy the issues arising in a timely manner?

For specifying our AI Ethics Life Cycle and deriving concrete measures while allowing for an agile approach to ethics, we are open and happy to collaborate with other parties. We strongly believe that the integration of ethics in the AI life cycle is crucial for realising high-quality, high-ethics AI, but also for the mitigation of risks on every level.

In the following weeks and months, we plan to engage in more open conversations about our framework, to organise workshops, to publish about our approach and to consult companies on how to implement those efficiently.


bottom of page