7. Conclusion

For several decades, AI has led to the development of operational applications in many fields, such as diagnostics and decision support with knowledge-based systems or neural networks, as well as pattern and speech recognition with stochastic models. The introduction of DNNs has made it possible to achieve unprecedented levels of performance, attracting unprecedented media coverage in virtually all sectors. As ever, transferring laboratory research to real-world applications requires substantial investment in time and money. There's a long way to go between an initial announcement and an actual marketed product. In medicine, in particular, the deployment of applications implies a complex evaluation mechanism, particularly in the clinical field. Finally, the success of an application implies genuine confidence in AI, as well as clear identification of responsibilities (cf. the autonomous...