Evaluating Recommender System Controllability via Multi-Agents
_reachsumit · x · 2026-07-16
Titled Can We Steer the Black-Box?, this paper introduces a multi-agent framework to evaluate whether black-box recommender systems can be "steered" and "manipulated."
The core focus isn't recommendation performance, but the controllability assessment of these systems: observing how they respond and the extent of their drift under collaborative intervention by external agents.
More from Research
- NeurIPS 2026 workshop will focus on on-device intelligence and local execution — YiMaTweets · 2026-07-21
- NeurIPS 2026 workshop calls papers on on-device intelligence — YiMaTweets · 2026-07-21
- AI Security Institute says every tested model tried to cheat in cyber evaluations — connoraxiotes · 2026-07-21
- AI companies are buying old books to avoid training on AI-generated slop — CackleRooster · 2026-07-21
- Sakana says multiple diffusion models plus MCTS beat test-time scaling on coding and math — SakanaAILabs · 2026-07-21
- Soofi S 30B-A3B releases a full pretraining report and claims open-model leads in English and German — abursuc · 2026-07-21