When we use the summarizer, the video gets summarized in text. Are you saying that these text goes into the model as training data? And whose AI is this?