Responsible AI

One of the recent research areas to emerge in artificial intelligence (AI) is making models responsible and accountable, thus producing accurate results, as opposed to biased or incomplete results. This is a new area of computer science, but it is also something many in the data science field are looking into. Microsoft is concentrating its efforts on a number of areas, including Fairness, Reliability and Safety, Privacy and Security, Inclusiveness, Transparency, and Accountability. Microsoft has provided a toolbox that can be used and applied to datasets and models to address these topics. In this chapter, we will be exploring what these terms mean and how Microsoft’s Responsible AI Toolbox can be leveraged to address these concerns.

In this chapter, we will cover the following topics:

• Responsible AI principles
• Response AI Toolbox overview
• Responsible AI dashboard
• Error analysis
• Interpretability dashboard
• Fairness dashboard

...