arXiv Computation and Language

Rating the Pitch, Not the Product: User Evaluations of LLMs Reflect Expectations More Than Performance

arXiv AI
3d ago

When Both Sides Adapt: The Challenges of Evaluating Interactive LLM Systems

The paper investigates how user adaptivity in interactive LLM systems affects evaluation outcomes. It distinguishes between the common‑distribution effect (model performance on a fixed prompt set) and the induced‑distribution effect (performance on prompts generated by user interaction). The authors show that offline benchmarks and live A/B tests target different estimands, can rank models differently, and that user adaptivity can increase estimation variance and lead to challenges such as reward hacking and novelty effects.

By Hannah Li
Hugging Face Trending Papers
Jun 28

Cognitive World Models for Process-Level Social Influence Evaluation

Social influence dialogue changes user behavior by altering internal cognitive states. The central evaluation question is whether the user's beliefs, desires, intentions, and emotions measurably change over the course of conversation, a process-oriented criterion that neither surface-level text metrics (BLEU/ROUGE) nor single-score LLM judgments can capture.

arXiv Computation and Language
Oct 28

Words or Deeds? How LLMs Predict Users' Decisions in Human-Agent Dialogue

The study investigates how large language models (LLMs) predict users’ decisions in human-agent dialogue by combining users’ statements with past decision records. Using 1,260 rounds of a repeated prisoner's dilemma with 42 users and five agents, the researchers found that decision records alone yielded strong predictive performance (AUC = 0.80), while adding dialogue did not improve overall accuracy but did increase agreement with users’ statements. However, predictions over‑emphasized statements, especially when users had not yet made a decision, and failed to capture variations in cooperation that depended on the specific agent or users’ trait alignment.

By Baiqiao Zhang, Shuai Ma, Zhifeng Liao, Chao Zhou, Jianshu Wang, Yuhan Zhao, Chaoyue Zhang, Tongjun Si, Yuying Tang, Xiangxian Li, Yong-Jin Liu, Yulong Bian
Hugging Face Trending Papers
Oct 27

Words or Deeds? How LLMs Predict Users' Decisions in Human-Agent Dialogue

The study examines how large language models (LLMs) predict users’ decisions in human‑agent dialogue by combining users’ statements with past decision records. Using 1,260 rounds of a repeated prisoner's dilemma with 42 users and five agents, the researchers found that decision records alone yielded strong predictive accuracy (AUC = .80), while adding dialogue did not improve overall performance but did increase agreement with users’ statements. However, predictions over‑emphasized statements, especially when users had not yet made a decision, and failed to capture variations in cooperation that depended on the specific agent or users’ trait alignment.

arXiv Computation and Language
Oct 23

Abstract Priorities, Concrete Rewards: How Stated Preferences Relate to What People Reward in AI Responses

The study investigates how users’ stated preferences for AI-generated responses align with the actual features they reward. Using the PRISM dataset, researchers compared participants’ reported priorities for seven response attributes with their ratings of over 29,000 responses from 21 large language models. They found that attributes not explicitly surveyed—such as length, formatting, and refusal language—were as strongly linked to ratings as the surveyed ones, and that written feedback was a better predictor of how users weighted these concrete features than their stated priorities.

By Zihan Chen, Lei Nico Zheng, Di Zhu
arXiv AI
Oct 26

Validating the Comparison, Not Just the Scorer: Why Overall Accuracy Can Mislead in Interactive LLM Evaluation

The paper argues that relying solely on overall accuracy or F1 scores to validate automatic scorers in LLM evaluation can conceal prompt‑specific errors that distort comparative results. By auditing six models on 180 invented‑entity questions across six prompts, the authors show that a scorer’s overall F1 of .916 masks substantial variation—up to 33.3%—in error rates depending on the prompt. In a second audit, differences in measured prompt effects dropped dramatically when multiple raters were used, highlighting the need to validate the exact comparison rather than the scorer alone.

By Ancil Crayton
arXiv AI
Sep 24

Ask Which, Not How Good: Sizing Benchmarks Scored by an LLM

The study analyzes 373,019 judgments from LLM‑scored benchmarks, decomposing variance into system, item, judge, and interaction components via generalizability theory. It finds that with a single judge, generalizability converges to a ceiling determined by the system‑by‑judge variance, which is substantially lower in pairwise preference settings, allowing one judge to suffice. The research also reveals significant biases in presentation order and highlights that many published win‑rate claims fall below the measured floor of the benchmarks.

By Atul Anand
arXiv Machine Learning
Aug 14

A Probe Direction Is a Property of Its Prompt

arXiv:2608. 13329v1 Announce Type: new Abstract: A model that behaves differently when it senses it is being tested would undermine the evaluations we rely on, so recent work has sought to read that sense directly from a model's activations.

By Valentin No\"el