This machine learning, which the developers call C-3PO (heh), worked by analyzing a person's browsing history, shopping history and financial details. The data was provided by Leopard Mobile, a Taiwan-based internet company. The neural network then analyzed the pop-up notifications people were getting and which ones they clicked on. As a result, the AI was able to make push notifications "smarter," reducing the number of overall notifications and increasing the click through rates on the ones that did appear, according to the article.
The team still has work to do. The next steps are to improve the neural network model by decreasing the number of complex tasks it has to perform. Additionally, they would like to apply this model to a system that advertisers could use to optimize when and how often they are delivering ads.
Anyone who's struggled with the sheer number of push notifications on their phones can see why this could potentially be a good thing. The easiest solution to decreasing push notifications is to turn off notifications or delete an app altogether. This gives users and app developers both a potentially better option.