Technosciences such as biotechnologies aim to analyse, modify and reconstruct livings and complex biochemical components, using top-down or bottom-up approaches to simplify these complex systems. Because of the large number of involved parameters, these technosciences need more and more big data techniques (Artificial Intelligence with machine and deep learning) to find efficient correlations to predict the evolution of systems in their environments. This prediction is especially used to reconstruct livings or artificial pieces of livings in biotechnology and to give medical diagnostics in medicine. Two examples corresponding to these two different cases are analysed in this paper. The aim is to highlight interests and limitations of such digital approach in terms of human free will and responsibility, in relation to provisional scientific truth. A difference is established between classical digital-assisted technoscience and digital-driven technoscience through an epistemological analysis of this new way to practice science. This allows to precise values which are favoured in a digital driven science, what we call an « ethics of knowledge ». Thus “capabilities to predict for making “, based on efficient correlations, look more important than “capabilities to know for making » and determine complex biological mechanisms or causes of evolutions. Consequences about technoscientific mentality are drawn and practical ethical questioning is consequently reassessed.