feet on a translucent ceiling
PHOTO: John Robert Marasigan

Artificial intelligence (AI) is making advertising more effective, but it’s also making it more opaque. With each new layer of automation, human beings are moved one step further from the most crucial marketing decisions of any digital campaign. After all, if AI makes its own decisions about which audiences to target, how can marketing professionals be confident that they’re reaching the right people?

AI systems were once primarily focused on efficient bidding, but new innovations have expanded their capacity to include audience segmentation, creative optimization and ad unit design. Until brands and agencies have ways of validating AI decision-making, it will be difficult for stakeholders to trust new technologies. Advertisers will be hesitant to adopt artificial intelligence practices unless vendors develop transparency solutions suited to new, automated technology.

In short, it won’t be enough for AI technologies deliver great results — they will also need to show their work.

The Evolution of Transparency in an Automated Era

In some ways, the AI explosion represents the latest challenge for transparency advocates as they strive to keep up with new technology.

As programmatic advertising has grown to become the dominant method of purchasing digital media, our industry has worked diligently to adapt. Over the past several years, advertisers, publishers and vendors have coalesced around solutions like Ads.txt, from the Interactive Advertising Bureau Tech Lab, that empower marketers to understand where their ad dollars are going. Marketers have also gained the ability to see which sites their ads run on, determine how much money goes to each of technology partner and verify whether the intended user actually saw the ad.

Now that marketers can gain insight into what their ad campaigns are doing, it’s up to vendors to start helping them understand why.

Today, advertisers select target audiences and use automation to evaluate and reach them. As AI becomes more powerful, advertisers will use algorithms powered by first- and third-party data to make audience decisions and selections on their behalf. In turn, vendors will need to provide advertisers with the means to understand these audience decisions; they will need to have systems in place that clarify the decision-making process for marketers. The same will hold true for all automated decision-making processes — those currently in use, like AI processes used for ad optimization, and all automated processes to come.

Related Article: Artificial Intelligence: Taking the Good With the Bad

Finding the Right Level of Transparency

For marketers and vendors, the questions then become “How much transparency is needed?” and “What does this transparency look like?”

In recent years, some marketers have tried to achieve transparency by obtaining impression-level campaign detail from their vendors. But more often than not, the sheer volume of data is overwhelming and, in some cases, it creates more questions than answers. Other times, marketers may identify high-level campaign trends, but that information doesn’t translate to meaningful insight or provide the understanding needed to build trust.

Ultimately, it will be up to technologists to build transparency functions within artificial intelligence products. Today, advertising technology vendors are using data science to make smarter marketing decisions. In the future, those vendors will also need to use data science to present decision-making information to marketers in a way that is easy to understand and meaningful.

Related Article: Brands Still Haven't Tapped AI's Full Promise

The AI Revolution Is Upon Us: Let's Make It Work for Our Industry

AI is set to command numerous critical marketing functions in the near future, so brands and agencies should start demanding real transparency from vendors now.

In five years or so, marketers won’t be buying audience segments — they’ll be buying data science products that autonomously construct audience segments. To establish a quality marketplace, vendors and marketers will need to work together to create standards for exposing AI decision-making logic in a way that allows marketers to meaningfully understand how a given AI brain “thinks.” From there, they’ll be able to understand the value and validity of any given AI platform, and choose the right technology for their campaigns.

By taking proactive steps toward transparency today, we can begin laying the groundwork for the AI ecosystem we want to inhabit several years down the road. If we shed light on how our new technology tools work, we’ll be able to create an industry where AI and data science can successfully make our marketers smarter. If we don’t, we’ll only be adding more ambiguity and uncertainty to a space that is already too opaque.