IBM working on AI tools to fight bias in online ads

Seeking to address discrimination concerns, IBM is working on new artificial intelligence (AI) tools that would make sure online advertising algorithms don’t unfairly exclude women and minorities.

Researchers and civil rights groups have found some online audiences, including black people and women, can be excluded from seeing employment listings, housing ads, and other ads due to automated advertising strategies.

Following complaints by federal regulators and activists, Google and Facebook, the world’s largest sellers of online ads, have enacted changes to reverse this trend.

Now, IBM is launching a new AI project to combat this problem, according to Reuters.

IBM says a team of 14 employees will research “fairness” in online ads over the rest of 2021. They will explore strategies for spotting and mitigating unintended bias in digital advertising, whether in the targeted audiences or the ads themselves.

IBM is launching the new initiative as ongoing concerns about data privacy reshape internet marketing.

“The foundation of advertising is crumbling and we have to rebuild the house,” IBM senior vice president Bob Lord told Reuters. “While we’re at it, let’s ensure fairness is in the blueprint.”

IBM sells ads on its online properties, including The Weather Company. It also markets ad software to other businesses and could eventually pitch clients on anti-bias AI systems.

IBM’s initial advertising analysis suggests online ads can be shown fairly equally to all groups without affecting factors, such as the percentage of users who click an ad, Robert Redmond, IBM’s head of AI ad product design, said.

Among other tasks, IBM’s team plans to analyze data from the nonprofit Ad Council’s recent public service announcements about COVID-19 vaccines.

“Systemic bias is unfortunately baked into nearly every corner of the advertising industry,” Ad Council President Lisa Sherman told Reuters.