Agenda Numerev : « Human and/or Machine Learning? » - Dominique Boullier
![](/sites/default/files/2020-11/Seminaires.jpg)
The promises of Machine Learning are too often oversold but what is right, is the challenge they bring in about the division of learning , as Zuboff calls it. If Machine Learners go on in the direction of disembedding themselves from organizations' design and experts' contributions, they will face major setbacks by overestimating the value of their predictions. But at the same time they will gain so much advantage in terms of deep and precise knowledge of social behavior that they will put their companies in position of bypassing the legal and institutional bodies, such as governments and cities. Another path is still open to develop a "human-machine learning" that would empower organisations and people, not only at the moment of using the outputs of ML but at the very core of the programmation activity.
Dominique Bouiller is professor of universities in sociology, researcher at Centre d'Études Européennes et de Politique Compérée at Science Po Paris. He specialises in the usage of digital and cognitive technologies and is the author of many books on the topic.