The IBM toolkit promises to reduce advertising distortion

There are many ways that bias can be woven into artificial intelligence (AI) and machine learning (ML) algorithms and decisions, despite best efforts to identify and eradicate it. It can be buried in the data used to generate the algorithms, in the training process itself, or arise from how the algorithms are used to make decisions.

In 2018, IBM launched AI Fairness 360, an open-source toolkit for checking and mitigating bias in datasets and ML models, and later added support for measuring uncertainty. The tool has improved the fairness of home loan, insurance, and medical decisions.

IBM’s new open-source Advertising Toolkit for AI Fairness hopes to do the same for the advertising industry. Consumers do not struggle to agree to better ads in the same way that they struggle to agree to a better mortgage rate or medical procedure.

“The real core of all of this is integrating tools to detect and mitigate bias with core marketing and advertising technologies,” Bob Lord, IBM senior vice president of the Weather Company and Alliances, told VentureBeat.

Statista estimates that companies spent $764 billion on advertising in 2021 and expects that number to rise to over $1 trillion in 2026. Better detection and mitigation of bias could help businesses, nonprofits, and governments get more value from their ad spend to different groups. It can also help improve the social determinants of health.

Advertising meets technology

“The bias that exists in advertising is historically ingrained in our marketing,” Lord said. It starts with how data scientists model segment data and model consumers. Now the advertising industry is going through a convergence of marketing and technology. “We’ve gotten really good at targeting people in advertising,” Lord said. But when targeting people with new ML algorithms, advertisers have also sub-optimized results for specific groups.

For example, IBM worked with the Ad Council on a project to understand the impact of bias in an algorithm-driven COVID-19 vaccine awareness campaign. The system dynamically delivered over 10 million ad impressions composed of 108 different creative variations selected by the algorithms. Over time, the system optimized ads for women aged 45-65, who ended up clicking through 32 times more than average.

While this might have been a great result for a new handbag accessory, it was sub-optimal for raising awareness of COVID-19 for other demographics. “The bias is not intentional,” Lord explained. “It’s hidden in the technology, and we don’t see it because we haven’t built preload detection technology into the machines yet.”

Lord’s team has already integrated this technology into AI and ML development workflows for mortgage applications and insurance underwriting. Today they work with some quick service companies to analyze marketing campaigns after the fact. “I’m hoping that in a year from now we’ll be able to incorporate this technology from scratch,” Lord said.