Controllability

In the context of AI, controllability refers to the ability to control the output of a model. This is a key area of research in AI safety, as it is important to be able to ensure that AI systems behave in a way that is aligned with human values.

Controllability is a key component of any successful AI application, as it can have a significant impact on the quality of the model's output.