• by Laurie Sullivan , Staff Writer @lauriesullivan, February 8, 2021

Google’s announcement validates the importance of using artificial intelligence (AI) to power the advertising ecosystem, but the industry must remain open to collaborate with all, said Sheri Bachstein, global head of The Weather Company and IBM Watson Advertising.

Bachstein referenced Google’s tests with its new API called Federated Learning of Cohorts that acts as a replacement signal for third-party cookies and the API that exists as a browser extension within Chrome.

IBM is betting on AI to become the backbone of the cookieless industry by bringing the technology into media and marketing. The processes must make sense of the data and offer insights for marketers, as well as maintaining consumer privacy.  

While IBM provides open technology to protect the open web, Bachstein wonders whether Google is putting up higher walls to protect its walled garden.

Interoperability will become one of the most important factors as the industry moves forward with the latest changes.

“If we build technology independent of one another it serves the purpose of one company, but not the others,” Bachstein said. “How do we transform the industry when there’s so much chaos? It’s something we should all push.”

Standards should exist, as media and marketing move to use more AI and automation. Creators must include a diverse group of people. Inherently, as developers develop technology, their own bias can infiltrate the platforms and the processes. Companies with a history of AI must provide that tech to the open marketplace.

AI continually learns, so the team must continually monitor what it learns. This is how companies root out bias.

IBM does this at The Weather Company. “Technologists create the platform, but then we have a team of meteorologists looking over the results,” she said. “They root out certain bias that might pop up in the forecast.”

When I asked what type of bias might surface in weather, Bachstein pointed to algorithm bias in terms of targeting and what ads should serve up in specific circumstances. She said there are many anomalies that occur in weather, such as precipitation. There are many micro-clients to consider.

“You can’t create AI and then walk away,” she said. “You need to ensure the companies you work with have a team behind the tech to continually look at the changes.”

Bachstein also spoke about Accelerator, IBM’s AI technology that dynamically assembles advertisements. From 20 assets, Accelerator can create between 80 and 100 ad units and serve them in real-time, similar to Google’s Dynamic Search Ads.