- Researchers have created AdGazer, an AI model trained on eye-tracking data to predict human attention to online ads with 83% accuracy.
- Page context drives up to one-third of an advertisement’s attention outcomes, challenging the assumption that creative content alone is most important.
- The academic demo, called Gazer 1.0, could be rapidly deployed in real-world ad-tech applications within months, not years.
Academic researchers from the University of Maryland and Tilburg University have unveiled a new AI tool that can forecast whether internet users will look at an online advertisement. This system, named AdGazer, leverages extensive historical eye-tracking data from real human viewers to make its predictions.
The model was trained using data from 3,531 digital display ads where people wore eye-tracking equipment. Consequently, it achieved a 0.83 correlation with actual human gaze patterns when tested on unseen advertisements. This high accuracy stems from its unique analysis of both the ad and the surrounding webpage content, unlike other tools that focus solely on the advertisement itself.
According to the study published in the Journal of Marketing, page context accounts for at least one-third of ad attention and 20% of brand-specific viewing time. This finding is significant for the marketing industry, which has long prioritized creative elements. The AI uses a multimodal large language model to assess semantic alignment between ad and page topics.
The team also developed an interface called Gazer 1.0 for public experimentation. This demo allows users to upload an ad and receive a predicted gaze duration and attention heatmap. Meanwhile, major tech firms like Perplexity and Anthropic are exploring alternatives to traditional display ads. However, this academic tool could quickly evolve into a commercial ad-tech product.
✅ Follow BITNEWSBOT on Telegram, Facebook, LinkedIn, X.com, and Google News for instant updates.
