There is a common consensus that state-of-the-art Artificial Intelligence (AI) and Machine Learning (ML) algorithms are powerful in terms of their accuracy, but they are also perceived as opaque not being transparent about how they arrive at their decisions.  This prevents the adoption of these powerful algorithms in Data-Driven Decision-Making. Even when in place, they can have a detrimental impact on the citizen, and there are well-documented examples of discriminatory outcomes in high-stakes algorithmic decision-making. Therefore, there is an urgent need to strike a balance between three goals, namely, accuracy, explainability and fairness.

In this talk, we will navigate through some of the latest advances in Mathematical Modelling and Optimization to enhance the transparency and fairness of ML algorithms. We will first focus on the training of ML models that trade off accuracy, explainability and fairness. Then, we will focus on the task of providing explanations to an existing ML model by means of the burgeoning field of Counterfactual Analysis.