Home » Tech » Can we trust algorithms?

Can we trust algorithms?

In a world that is becoming more computerized every day and where the use of personal digital data raises the question of our privacy, the word “algorithm” is making a strong entry in our vocabulary. It would have the announced goal of bringing artificial intelligence to the substance necessary for its programs so that it helps us better take charge of our lives. There are usually few problems in everyday life because we can judge the results obtained. The same is not true in the health field because we would like to understand what these algorithms are and what we can expect from them.

→ READ. Algorithms and artificial intelligence – a call to be undertaken in the hope!

Mathematicians, computer scientists and others tell us that an algorithm is a process made up of a set of given operations and rules to perform a precise calculation. To try to trivialize, they indicate that a cooking recipe that defines the ingredients and the precise way of using them for a culinary success behaves like an algorithm. Is ! But in this case we know the ingredients well and master each step. This is not enough to reassure us, especially when artificial intelligence and its algorithms invade medicine. In fact, it is now possible to collect all possible data on anyone. This data can be manipulated at will to create an algorithm that can have significant effects on people’s lives, for good or for bad, without any safeguards to be reassured.

“It is now possible to collect all the possible data on anyone”

It follows therefore mixed reactions close to absolute mistrust. However, this distrust is unfortunate because we really need algorithms in our modern lives. It is therefore essential to improve their transparency. Some, like the British mathematician Hannah Fry (1), argue more and more forcefully for the creation of a body which would control their marketing, especially when they are for medical purposes. In reality, we find ourselves with the algorithms in a situation comparable to that which we knew with the drugs about thirty years ago. One of the primary concerns was to differentiate between molecules that are beneficial in therapeutic terms and molecules that are uncertain, even dangerous. This situation logically led to the creation of an agency responsible for ensuring the good quality of medicines, today the National Medicines Safety Agency (ANSM).

See also  "Abimagic", time maker - Culture / Next

On the same model, an algorithm security agency could both protect their intellectual property and ensure beneficial effects for patients and society. It is she who would give the authorization to operate. This provision could be introduced into the text of the bioethics law for revision before the Parliament, unless taking proven risks because no one controls or labels the algorithms used today. The ideal would no doubt be for the European Union to ensure its sovereignty in this matter as it expresses the will. This is one of the ethical issues that I think are the most urgent to resolve.

(1) Author of Hello World. How to Be Human in the Age of the Machine, Doubleday, 2018.


Leave a Comment