From the course: Building a Responsible AI Program: Context, Culture, Content, and Commitment

Unlock the full course today

Join today to access over 24,700 courses taught by industry experts.

Data done right

Data done right

From the course: Building a Responsible AI Program: Context, Culture, Content, and Commitment

Start my 1-month free trial Buy for my team

Data done right

“

- When we hear stories about AI gone wrong, we can often trace the roots back to an issue with data. Data is a key ingredient in an AI system, so it's not surprising that poor data and ill-conceived data practices can lead to a whole host of problems. This is a typical data life life cycle. I want to touch on the major points where AI ethical issues intersect with the data life cycle. So let's walk through this to examine some common ethical issues. The first stage in the data lifecycle is planning, and if your plan is to weaponize data, that is, to use data in ways that might be unfair or harmful, or justify some kind of discrimination, then your project is going off the ethical rails before you even get started. In other words, examine the purpose of the project for ethical concerns first. The collection phase is where decisions around how to acquire data are made, and this is where you can run into issues of inappropriate or lack of consent, as well as privacy concerns. Since many…

Contents