IBM tackles bias in ad tech

IBM has launched an Advertising Fairness Pledge to tackle the risk of bias in advertising technology, backed up by an open source toolkit to help agencies and advertisers identify and mitigate biases in discrete data sets.

Why it matters

A recent Salesforce survey found that 62% of consumers were concerned about bias in AI – up from 54% two years earlier – emphasizing the imperative for brands and agencies to better understand its impacts. And last week, a report from the US Federal Trade Commission warned against relying on AI to solve online problems, not least since doing so can introduce “additional harms” – including bias and discrimination. 

Takeaways

• The initiative is supported by agencies, advertisers and industry bodies, including Delta Air Lines , WPP, Mindshare, 4A’s, IAB and the Ad Council. 
• Bias is often unintentional, a result of human assumptions and judgments encoded into algorithms that can result in unfair targeting, exclusion of certain groups, and marketing campaign failures. 
• IBM’s Advertising Toolkit for AI Fairness 360 solution deploys 75 fairness metrics and 13 state-of-the-art algorithms to help organizations gain a better understanding of the presence and impact of bias on their ad campaigns, as well as the makeup of their audiences.
• With increasing consumer demand for transparency in how their data is used, marketers must look for new ways to remain effective. Tapping into alternative privacy-forward data sources, such as weather data, can be effective predictors of behavior that could also help rebuild trust with consumers.

 Key quote 

“Consumers rightly expect brands to use their information in a fair way and for the industry to tackle data bias collectively, which can ultimately result in increased engagement and commercial outcomes” – Mark Read, CEO, WPP.

Sourced from IBM, FTC