From scientific journals and under-desk workstations, AI is entering the world of large corporations today, replacing traditional solutions with self-optimizing models. Although under the hood these are still algorithms based on back-propagation of the gradient known from the previous century, their development on an industrial scale brings many new challenges. During the meeting, we will start with a reminder of the basics of neural networks to show what the biggest players on the market are struggling with today through interesting examples of their applications.