The Generative Turn in Search and Recommendation

Search and recommendation systems power web search, short-video feeds, music, news, advertising, and e-commerce. They are now undergoing their most significant paradigm shift in a decade: multi-stage discriminative pipelines are giving way to unified generative foundation models that directly produce items, queries, slates, and entire user trajectories.

This transition promises to reduce cascaded error propagation, improve hardware utilization, and unlock optimization objectives beyond next-click prediction. Realizing the paradigm in production, however, requires serving billions of users, reasoning over catalogs approaching billions of candidate items, and responding within strict latency budgets.

The workshop brings together researchers and industrial practitioners through invited talks and a moderated discussion around convergence, scaling laws, trillion-item efficiency, continuous learning, cold start, fairness, and multi-objective trade-offs.