NExT-Search: Rebuilding User Feedback Ecosystem for Generative AI Search
Sunhao Dai, Wenjie Wang, Liang Pang, Jun Xu, See-Kiong Ng, Ji-Rong Wen, Tat-Seng Chua
2025-05-21
Summary
This paper talks about NExT-Search, a new system designed to make AI-powered search engines smarter by collecting detailed feedback from users at every step of the search process.
What's the problem?
The problem is that most generative AI search engines don't get enough useful feedback from users to know exactly where they're making mistakes or how they can get better, which slows down their improvement and can lead to less helpful results.
What's the solution?
To solve this, the researchers introduced special modes called User Debug Mode and Shadow User Mode, which let users give specific feedback at different points in the search. This helps the AI learn from its mistakes and improve much faster and more accurately.
Why it matters?
This matters because it means search engines powered by AI can become more reliable and helpful, making it easier for everyone to find the information they need and trust the answers they get.
Abstract
NExT-Search seeks to enhance generative AI search by introducing fine-grained feedback mechanisms at each stage of the pipeline using User Debug Mode and Shadow User Mode, facilitating continuous improvement.