RETINA — New AI Tool Can Predict Your Online Shopping Choices Based on Eye Movement

Researchers have developed a new AI model that leverages real-time eye-tracking combined with a deep learning algorithm to accurately predict user behavior on simulated online shopping platforms.

Called Raw EyeTracking and Image Ncoder Architecture (RETINA), the algorithm is different from conventional eye-tracking software by incorporating the entire range of captured data rather than aggregated information. Unlike traditional methods that struggle to track specific eye movements, RETINA processes raw eye movement data for each eye separately, providing a more comprehensive analysis.

The architecture claims to predict consumer choices using raw eye movement data without the intermediate step of calculating fixations. By utilizing the complete time series of eye-tracking data from both eyes as input, RETINA outperforms other machine learning methods in predicting choices among four laptops. The analysis indicates that RETINA can predict outcomes well in advance, achieving a predictive accuracy of over 0.7 with just 5 seconds of data.

Michel Wedel, one of the study's authors, highlights the vast amount of data processed—several hundreds of thousands of data points and millions of parameters for both eyes. This algorithm streamlines the processing of eye movement data, opening up numerous applications beyond online shopping behavior prediction.

The potential applications of this technology extend beyond online retail. The researchers expect its use in guiding shopping experiences within the metaverse, leveraging built-in eye-tracking tools in virtual reality (VR) devices.

While traditional smartphones can also track eye movements, their practical applications have been limited. The strength of AI lies in its predictive capabilities based on comprehensive data analysis, offering a promising tool for various domains beyond online shopping.