In every language, there are common mistakes. In Spanish, for example, por and para or ser and estar get mixed up easily. It would be useful to have decks dedicated to those errors. To take the example of por and para, a deck containing 250 sentences that use por and 250 that use para would pretty much guarantee that the mistake is not made anymore or nearly never made. These sentences could be AI -generated using the vocabulary already in the known deck.