Khan describes the problem with human planners: „Most of the time the people who are in charge of the planning are choosing to be on the upper side, as they do not want to be responsible for not having the components ready to deliver to the customers.“
To show how well the RNN (Recurrent Neural Networks) are already working Khan picks a component that has been calculated by the RNN: Compared with the human forecast which differs around 35-40%, the same forecast done by the AI model was only off by 18-19% in comparison to the real customers demand.
Currently for more than 75% of the examined components, the team of Khan was able to bring down the cumulative error.
„We are still at a very R&D level. So I currently see this as an additional tool to assist the humans.“ By gradually proving the results Khan hopes to establish RNN further in the supply chain.