PulseAugur
EN
LIVE 06:46:58

LLMs fail to replicate individual human responses, study finds

By PulseAugur Editorial · [1 sources] ·

A new research paper titled "Item-Mean Surrogates: Why Richer Persona Data Fail to Improve LLMs as Human Surrogates" has been published on arXiv. The study found that while Large Language Models (LLMs) can accurately predict average human responses to survey items, they fail to capture individual-specific variations. Even with richer persona data and fine-tuning, LLMs could only explain a small fraction of the respondent-specific variance, significantly underperforming human test-retest reliability. The research highlights that current LLMs exhibit "item-mean surrogacy," meaning they approximate item averages but not the nuanced, individual deviations required to truly substitute for humans. AI

IMPACT LLMs can approximate average human responses but cannot yet capture individual-specific variations, limiting their use as human surrogates.

RANK_REASON Research paper published on arXiv detailing findings about LLM capabilities. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

LLMs fail to replicate individual human responses, study finds

How we ranked this

Signal score
27 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
Research paper published on arXiv detailing findings about LLM capabilities. [lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
paper, model release
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

Full methodology in our editorial standards.

COVERAGE [1]

  1. arXiv cs.CL TIER_1 English(EN) · Daehwan Ahn, Chengfeng Mao, Dokyun Lee ·

    Item-Mean Surrogates: Why Richer Persona Data Fail to Improve LLMs as Human Surrogates

    arXiv:2608.29455v1 Announce Type: new Abstract: LLMs are increasingly used as human surrogates, often on the premise that richer persona data could make them substitutes or exploratory tools for specific individuals. We test this premise across four datasets covering more than 40…

RELATED ENTITIES

RELATED TOPICS