Is Deep Learning ready to satisfy Industry needs?

The impact that Artificial Intelligence is having in modern society is undeniable. Many companies are now using AI to improve the throughput and automate their processes. But the challenge is that Artificial Intelligence is both a source of enthusiasm and skepticism for industries. The manuscript points out the main causes of skepticism giving at the same time some possible technical solutions to exploit at the best the potentialities of AI even in those conditions in which the data are imbalanced and the object classes are not well separated. This work also emphasizes the delicate relationship between artificial intelligence, researchers, and industries, and tries to give an overview of a possible trade-off between the two parties. The document ends up proposing an 'interpretable learning' approach that can be exploited as a common language between the two players. The desirable practice would be to make AI explainable, provable, and easily understandable by the companies.