Does Goodnotes use document data to train AI models?
We use document data in order to provide and optimize our AI features. Raw documents stored by Goodnotes are stripped of any personal data before being used for training.
What data does Goodnotes collect from your institution’s use of AI features?
We collect anonymized statistical data to track the use of our AI features. We also collect Human Override Data (aka AI mistakes) and any "change" made by a teacher to override any decision made by AI. This helps us improve the accuracy and identify mistakes in the AI model.
How has Goodnotes designed Smart Clusters in line with responsible AI?
Smart Clusters have been trained on a diverse and representative data set. For example, its handwriting recognition component encompasses a wide range of handwriting styles and languages. Smart Clusters’ ability to comprehend language is powered by Open AI’s GPT-4 model, which is trained on a large sample of 1 petabyte of data. Moreover, whenever using Smart Clusters, teachers are reminded that AI is capable of making mistakes and they should verify its output.