4. Conclusion

Artificial intelligence is today a powerful vector of innovation in fields as varied as basic science, engineering, medicine, agriculture and industry. It enables us to extract knowledge from massive data, design complex systems, optimize resource consumption or assist medical diagnosis with increasing precision. These transformations are based on a combination of advances in machine learning, computing power and data access, and are transforming working methods in most scientific and technical disciplines.

These impressive and undeniable performances should not, however, overshadow the profound challenges that widespread use of AI technologies entails: algorithms need to be as robust, explainable, safe, energy-efficient and fair in their decisions as possible. Algorithmic biases, data quality and representativeness, increasing energy consumption and the difficulty...