AI is taking on an important role in all areas of business – helping companies make critical decisions that affect people’s lives, such as who qualifies for a mortgage, how much an insurance plan will cost or how well a medical treatment will fare on certain populations. As I discussed in an article in Future of Sourcing, because of this responsibility, it’s more important than ever that biases aren’t inadvertently being built into AI algorithms.
What are some of the ways data professionals can avoid data bias when building an AI solution? Think of the following:
- Hire a diverse team. The majority of data scientists today are white males; according to a Harnham’s Diversity Report for the U.S. Data & Analytics Industry, and only 18% of data scientists are women. Without diversity, there is a possibility for unintentional racial and gender bias to creep into the data.
- Use diverse data. It’s important to constantly check the data for bias. Are mortgages being denied based on a person’s zip code, ethnic group or class? If so, make sure sufficient data is being added to provide a bigger picture.
- Be transparent. Always be prepared to disclose the type of data used to train an algorithm and the criteria that helped make decisions, such as who gets a mortgage, or how risk is assessed.
- Continuously test the algorithm. It’s critical to monitor your algorithm continually, not just to improve results, but also to make sure new data is not bringing new biases to the application.
AI is having a major impact on how decisions are being made, so data professionals must be continuously on the lookout for biased data that left unchecked can damage reputations, create legal implications and most importantly, decide the fate of individuals without sufficient data to support those decisions.