Top Stories
Delta, Kellogg join IBM pledge to fight bias in online ad targeting
Global Banking & Finance Review®
Company
By Paresh Dave
OAKLAND, Calif. (Reuters) – Delta Air Lines Inc, Kellogg Co and several big ad agencies on Monday pledged to counteract unwanted biases in the algorithms that power their online advertising.
The action, announced as part of an ad industry conference in Cannes, France, is one of the biggest steps taken to ensure stereotypes related to gender, race and other characteristics are not unfairly deciding to whom an ad is shown.
Growing use of automated software to buy ads, greater attention on diversity and the looming risk of scrutiny from consumer protection regulators has brought urgency to the issue.
The advertisers and agencies will use algorithms and guidance from IBM Corp to identify skews in audiences for ads and balance them out. IBM began testing the tools internally a year ago. They are free, but likely will require added staffing to deploy.
Advertisers acknowledge that tweaks could mean fewer clicks for some time, but that being more inclusive eventually will increase sales.
“Anything we can do to champion greater equity has huge business impact,” said Shannon Womack, director of U.S. brand marketing at Delta.
In a test, WPP Plc’s Mindshare agency found an unnamed goods company would direct online ads to mostly women because female buyers were more common. Mindshare determined fairness could be improved by six times on a measure of disparate impact by including more male viewers, with the margin of error on predictions related to buyers widening by under two percentage points.
“We can start to break the cycle of some long-held beliefs,” said Adam Gerhart, global chief executive of Mindshare, which counts IBM as a client. “The biggest determinate [of becoming a buyer] may not be your gender but your role in a household or other characteristics.”
Agencies Dentsu Group Inc and Publicis Groupe SA and ad tech companies Criteo SA and Magnite Inc also signed the pledge, IBM said.
(Reporting by Paresh Dave, Editing by Louise Heavens)
Algorithmic advertising refers to the use of automated systems and algorithms to purchase and display online advertisements, optimizing ad placements based on various data points.
Bias in advertising occurs when certain demographics are unfairly targeted or excluded based on characteristics like gender or race, leading to skewed representation in ad campaigns.
IBM provides tools and algorithms to help advertisers identify and mitigate biases in their advertising strategies, promoting more equitable ad targeting.
Audience targeting is the practice of identifying and reaching specific groups of consumers based on demographics, interests, and behaviors to optimize advertising effectiveness.
Diversity in advertising aims to represent various demographics fairly, which can enhance brand perception, broaden market reach, and ultimately lead to increased sales.
Explore more articles in the Top Stories category
