Artificial Intelligence (AI) is moving from science fiction to everyday use, but has no universal definition. AI potentially encompasses any task that would normally require human intelligence, which means it captures technologies such as machine learning, visual perception, speech recognition, decision-making, and language translation.(1)

Not only does AI cover a broad scope of technologies, its scope is also always evolving. In order to allow machines to learn and reach their full potential, large amounts of data are often required and therefore the considerations that apply to big data will often overlap with AI. This is where data protection comes into play, and where it borrows characteristics and challenges from big data. In many cases, the outcome of AI cannot be predicted and therefore, it is often difficult to pin down what the exact use cases will be at the point of collection, and what the nature of any data coming out of the application of AI will be. In addition, gathering and managing large datasets will pose its own challenges, including reaching out to data subjects if use cases change or if new information is gleaned from the application of AI.