PrefDisco: Benchmarking Proactive Personalized Reasoning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Shuyue Stella, Bose, Avinandan, Brahman, Faeze, Du, Simon Shaolei, Koh, Pang Wei, Fazel, Maryam, Tsvetkov, Yulia
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915834720419840
author Li, Shuyue Stella
Bose, Avinandan
Brahman, Faeze
Du, Simon Shaolei
Koh, Pang Wei
Fazel, Maryam
Tsvetkov, Yulia
author_facet Li, Shuyue Stella
Bose, Avinandan
Brahman, Faeze
Du, Simon Shaolei
Koh, Pang Wei
Fazel, Maryam
Tsvetkov, Yulia
contents Current large language model (LLM) development treats task-solving and preference-alignment as separate challenges, optimizing first for objective correctness, then for alignment to aggregated human preferences. This paradigm fails in human-facing applications where solving a problem correctly is insufficient if the response mismatches the user's needs. This challenge intensifies in just-in-time scenarios where no prior user interaction history exists due to cold-start conditions or privacy constraints. LLMs need to proactively identify what they don't know about the user, strategically elicit preference values through questioning, then adapt their reasoning processes and responses accordingly -- a complicated chain of cognitive processes which we term personalized reasoning. We introduce PrefDisco, an evaluation methodology that transforms static benchmarks into interactive personalization tasks using psychologically-grounded personas with sparse, context-dependent preferences, and define PrefAlign as a fine-grained rubric-based metric for measuring preference alignment. PrefDisco builds scenarios where identical questions require different reasoning chains depending on user context, as optimal explanation approaches vary by individual expertise and preferences while maintaining factual accuracy. Evaluation of 21 frontier models across 10 tasks reveals 29.0% of naive personalization attempts produce worse preference alignment than generic responses, yet generic responses also fail to serve individual user needs. These findings suggest personalized reasoning requires dedicated development rather than emerging naturally. PrefDisco provides a foundation for developing systems that can adapt to individual users in education, healthcare, and technical domains where personalization is critical.
format Preprint
id arxiv_https___arxiv_org_abs_2510_00177
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle PrefDisco: Benchmarking Proactive Personalized Reasoning
Li, Shuyue Stella
Bose, Avinandan
Brahman, Faeze
Du, Simon Shaolei
Koh, Pang Wei
Fazel, Maryam
Tsvetkov, Yulia
Computation and Language
Artificial Intelligence
Current large language model (LLM) development treats task-solving and preference-alignment as separate challenges, optimizing first for objective correctness, then for alignment to aggregated human preferences. This paradigm fails in human-facing applications where solving a problem correctly is insufficient if the response mismatches the user's needs. This challenge intensifies in just-in-time scenarios where no prior user interaction history exists due to cold-start conditions or privacy constraints. LLMs need to proactively identify what they don't know about the user, strategically elicit preference values through questioning, then adapt their reasoning processes and responses accordingly -- a complicated chain of cognitive processes which we term personalized reasoning. We introduce PrefDisco, an evaluation methodology that transforms static benchmarks into interactive personalization tasks using psychologically-grounded personas with sparse, context-dependent preferences, and define PrefAlign as a fine-grained rubric-based metric for measuring preference alignment. PrefDisco builds scenarios where identical questions require different reasoning chains depending on user context, as optimal explanation approaches vary by individual expertise and preferences while maintaining factual accuracy. Evaluation of 21 frontier models across 10 tasks reveals 29.0% of naive personalization attempts produce worse preference alignment than generic responses, yet generic responses also fail to serve individual user needs. These findings suggest personalized reasoning requires dedicated development rather than emerging naturally. PrefDisco provides a foundation for developing systems that can adapt to individual users in education, healthcare, and technical domains where personalization is critical.
title PrefDisco: Benchmarking Proactive Personalized Reasoning
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2510.00177