Spotify reframes recommender systems for the agentic web in weekly IR papers roundup
_reachsumit · x · 2026-09-20
Author published Vol. 174 of the "Top Information Retrieval Papers of the Week" newsletter, covering ten studies:
- Spotify's "Recommender Systems in the Agentic Web" argues that two decades of assumptions—humans directly receiving recommendations—break down as LLM agents browse, compare, and transact on users' behalf; it proposes a "delegation spectrum" framework placing decisions between fully human-led and fully agent-led based on three factors, including how explicitly preferences can be specified upfront and how verifiable outcomes are.
- Mixture-of-Experts LLMs as first-stage retrievers (Shrestha et al.)
- Adaptive randomized algorithms for item-to-item retrieval on billion-edge graphs (Jiang et al.)
- Deriving TF-IDF and BM25 from first principles via KL divergence (Ivan Silajev)
- Self-evolving memory for generative recommendation (Meta)
- Self-evolving search index (Lee et al.)
- Calibration in attention-based rerankers (Google)
- A fully open, efficient foundation model for math and agentic search (He et al.)
- Scaling Transformers for industrial recommendation via transferable generative pre-training (Alibaba)
- Causal compression of lifelong user histories for recommendation (Tencent)
More from AGI Musings
- Researcher: Alignment Community Should Debate the Strongest Arguments, Not the Loudest Critics — RichardMCNgo · 2026-09-20
- Researcher: Good Work Doesn't Speak for Itself, Marketing Does — A_K_Nain · 2026-09-20
- Stanford AI professor Yejin Choi to headline Korea Society ethics and common values lecture series — stanfordnlp · 2026-09-20
- Frontier Token Costs Make It Hard for Low-Purchasing-Power Countries to Compete — Paimaamu · 2026-09-20
- Debate: Freeze frontier AI now, or is that just regulatory capture? — GarrisonLovely · 2026-09-20
- Josh Gans' AI doomer joke: the species AI wipes out isn't humans, it's New Caledonian crows — joshgans · 2026-09-20