Yandex has introduced Sona, an advanced generative recommender system designed to replace the traditional recommendation cascade. This innovative model combines candidate generation and ranking into a single transformer, eliminating the need for hand-engineered features.
In a recent A/B test conducted on Yandex Music, the implementation of Sona led to an impressive 11.42% increase in user likes, showcasing its effectiveness in enhancing user engagement. This development marks a significant step forward in the field of AI-driven recommendations, potentially setting a new standard for other platforms.
The launch of Sona reflects Yandex’s commitment to advancing AI technologies and optimizing user experiences. By simplifying the recommendation process, Yandex aims to provide more relevant content to its users, thereby improving satisfaction and retention rates.
Compiled automatically by the Tech AI Newsdesk from public AI-news sources and summarised in our own words.
