Meta, the parent company of Facebook, has developed a new artificial intelligence (AI) model called Lattice to improve its advertising system. Lattice is designed to generate income for the company by predicting how well an ad will perform and deciding the best placement for it. The model merges data from different meta-services and applications to gain a comprehensive overview of all placement options, such as feed, stories, reels, clicks, video views and conversations. With increased efficiency, Meta can react more quickly to new developments in the market.
The development of Lattice is in response to the decreasing amount of user data available due to regulations and user awareness of the value of their data. Apple’s query when installing an app has led to more people opting out of the transfer, causing a disagreement between Apple and Meta. To overcome this challenge, Lattice merges smaller AI models to make better predictions about ad performance with less user data.
Lattice is based on a transformer architecture and has shown an eight percent increase in ad quality in initial tests at Instagram. The Interactive Advertising Bureau (IAB) also sees Lattice as a key way of intercepting the effects of Apple’s query. Overall, Lattice’s comprehensive evaluation of ad placement options and success factors for advertisers makes it a powerful tool for Meta’s advertising system.
