As AI systems are increasingly deployed into our society, they raise several concerns, notably in terms of privacy, biases, unexplainable decisions, and non-aligned objectives. In this presentation, I specifically focus on Machine Learning and explain what are those concerns, along with a few examples, and what are some solutions. These solutions include technical advances, but most importantly the need for human reflection within the design of such systems.
A talk that I gave at a seminar of the Abylsen company. The slides are in French.