Language Models Don't Know What You Want: Evaluating Personalization in Deep Research Needs Real Users

TL;DR

This paper introduces MyScholarQA, a personalized deep research system that infers user interests, proposes tailored actions, and generates multi-section reports, emphasizing the importance of real user feedback for genuine personalization.

Contribution

The paper presents MyScholarQA, a novel personalized research assistant that combines synthetic benchmarking with real user studies to uncover nuanced personalization errors.

Findings

MySQA outperforms baselines in citation metrics and personalized actions in synthetic tests.

Real user interviews reveal nine nuanced personalization errors undetectable by LLM judges.

Qualitative feedback provides lessons for future personalized research system design.

Abstract

Deep Research (DR) systems help researchers cope with ballooning publishing counts. Such tools synthesize scientific papers to answer research queries, but lack understanding of their users. We address this with MyScholarQA (MySQA), a personalized DR agent that: 1) infers a profile with a user's research interests; 2) proposes personalized actions for a user's input query; and 3) writes a multi-section report for the query that follows user-approved actions. We first test MySQA with NLP's standard protocol: we build a benchmark with synthetic users and LLM judges, where MySQA beats baselines in citation metrics and personalized action-following. However, we suspect this process does not cover all aspects of personalized DR users value, so we interview users in an online version of MySQA to unmask them. We reveal nine nuanced errors of personalized DR undetectable by our LLM judges, and we study qualitative feedback to form lessons for future DR design. In all, we argue for a pillar of personalization that easy-to-use LLM judges can lead NLP to overlook: real progress in personalization is only possible with real users.