A misleading and controversial topic about AI. When an AI model is said to be transparent, it means that there is documentation regarding its design and purpose. It also means that certain outcomes can be explained and that users understand the limitations within an AI model.
Transparency is important if a model is expected to be compliant with regulations. However, when it comes to chatbots and deep learning state-of-the-art methods, true transparency is almost impossible because of the complexity of the AI's architecture (see
neural network).