Is your media plan biased? New, free AI Toolkit analyzes campaigns

Marketers curious about the bias in their campaign targeting can put their media plans to the test by running them through a free open source toolkit from IBM.

The idea behind using the tool, called the Advertising Toolkit for AI Fairness 360, is to stay one step ahead of regulation by ensuring that the advertising industry eradicates biases in campaign planning, especially the invisible kind that can be reinforced. by relying on data segments (even a marketer’s first-party data) and black box algorithms.

IBM is looking to drive greater involvement from the advertising industry as many companies make commitments to promote greater inclusiveness in campaign targeting and messaging, IBM SVP Bob Lord said.

IBM also wants to enable data sharing between partners for more accurate research into the effects of bias and more effective training of machine learning models, Lord said.

The IAB, which plans to evangelize the initiative, expects the toolkit to be a valuable resource for advertisers at a time when consumers want to support brands that conduct business responsibly and use technology, said IAB EVP and Chief Industry Growth Officer Sheryl Goldstein.

“As an industry, we need to ask ourselves whether we are accountable in the way we create media and the way we segment and target consumers,” Goldstein said.

The IAB also sees industry-led efforts to reduce bias as an effective way to achieve self-regulation, as widespread bias in ad technology is likely to lead to government action if not proactively addressed, Goldstein added.

IBM’s announcement of the initiative comes at an important time for the advertising industry, said Adam Gerhart, CEO of Mindshare Global.

He identified two emerging trends that require an industry-wide approach to reduce AI bias: a greater appreciation of advertising’s impact on the world and its ability to shape perceptions, plus the accelerated pace of data proliferation and automated use of data in advertising applications.

“If we don’t get those two things sorted out, the chance for bias to persist and even accelerate in the advertising industry is unchecked,” Gerhart said.

Mindshare has started using the toolkit in its clients’ campaigns to identify bias in the data and algorithms commonly used in the industry, mitigate the effects of that bias, and validate the impact that throttling has on performance Gerhart said. Mindshare performs this analysis after the campaign.

In addition to Mindshare, Delta Air Lines, WPP, the 4As, the IAB and the Ad Council have committed themselves to the initiative.

One type of targeting bias that the toolkit should weed out is bias caused by propensity models trained using historical first-party data, Gerhart said. By reducing the preference for targeting audiences who have completed transactions in the past, new audiences can open up to campaign targeting that would have been ignored by models looking only at the status quo.

How Bias Detection Works

To identify and reduce biases ingrained in the datasets advertisers and publishers use for ad targeting and campaign planning, the toolkit uses 75 fairness metrics and 13 algorithms derived from IBM Watson Advertising’s previous research on the prevalence of bias in AI.

For example, one of the toolkit’s algorithms focuses on limiting campaign targeting’s reliance on protected variables such as race and gender.

As more partners use the toolkit for their own efforts to reduce bias, that data will further train the AI ​​powering the software.

Once advertisers have gathered all the data they want to use for a campaign, they can run those details through the toolkit. In the toolkit, for example, it can be found that a cleaning products campaign was biased towards women who have purchased these products in the past. The brand may then decide to target more men to reduce the bias.

The toolkit does not fully automate the reduction of bias, but reveals blind spots in marketers’ existing campaign parameters. After analysis, marketers can make smarter decisions about which audiences to target and how best to reach those audiences, IBM said.

While the toolkit analyzes bias in campaign planning alone to begin with, IBM plans to develop capabilities in the future to reduce creative bias.

Leave a Comment

Your email address will not be published.