AI technologies provide user-friendly solutions at a scale and efficiency that was not imaginable before. In decision making, AI can help to eliminate human bias, and to make wise decisions that benefit human beings and societies. Its many benefits have caused the AI revolution to have a huge impact on all aspects of modern human life. As AI is fusing into our lives, its potential harms have become more evident. We all have perhaps faced or heard many of these concerns. The concept of Responsible AI has been introduced to minimize the drawbacks of AI.
It is known that AI is as good as the data it is built on. When data does not contain enough signals to address business needs, no model can achieve a high-enough performance to address those needs, hence, responsible AI requires that responsible data be collected. Since responsible data is often scattered across multiple sources, responsible data integration is required for collecting the responsible data. Consider the following example in the healthcare domain.
Responsible AI introduces new challenges and requirements for data integration, that require revisiting different tasks in the pipeline of data integration to make sure these needs are satisfied.