# The Inadequacy of Offline LLM Evaluations: A Need to Account for Personalization in Model Behavior

arXiv: https://arxiv.org/abs/2509.19364 · HTML: https://mlsanity.com/paper/2509.19364/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- Builds on 29 of its 37 references that are in the map.
- First posted on arXiv 18 Sep 2025.

## Builds on (most cited first, 29 of 29)

- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [A Survey on Evaluation of Large Language Models](https://mlsanity.com/paper/2307.03109/) (2023)
- [Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference](https://mlsanity.com/paper/2403.04132/) (2024)
- [Personalizing Dialogue Agents: I have a dog, do you have pets too?](https://mlsanity.com/paper/1801.07243/) (2018)
- [Aligning AI With Shared Human Values](https://mlsanity.com/paper/2008.02275/) (2020)
- [WildChat: 1M ChatGPT Interaction Logs in the Wild](https://mlsanity.com/paper/2405.01470/) (2024)
- [From Crowdsourced Data to High-Quality Benchmarks: Arena-Hard and BenchBuilder Pipeline](https://mlsanity.com/paper/2406.11939/) (2024)
- [Scaling Synthetic Data Creation with 1,000,000,000 Personas](https://mlsanity.com/paper/2406.20094/) (2024)
- [AI and the Everything in the Whole Wide World Benchmark](https://mlsanity.com/paper/2111.15366/) (2021)
- [Large Language Models Sensitivity to The Order of Options in Multiple-Choice Questions](https://mlsanity.com/paper/2308.11483/) (2023)
- [MT-Bench-101: A Fine-Grained Benchmark for Evaluating Large Language Models in Multi-Turn Dialogues](https://mlsanity.com/paper/2402.14762/) (2024)
- [WildBench: Benchmarking LLMs with Challenging Tasks from Real Users in the Wild](https://mlsanity.com/paper/2406.04770/) (2024)
- [Two Tales of Persona in LLMs: A Survey of Role-Playing and Personalization](https://mlsanity.com/paper/2406.01171/) (2024)
- [When Benchmarks are Targets: Revealing the Sensitivity of Large Language Model Leaderboards](https://mlsanity.com/paper/2402.01781/) (2024)
- [Position: Evaluating Generative AI Systems Is a Social Science Measurement Challenge](https://mlsanity.com/paper/2502.00561/) (2025)
- [Toward an Evaluation Science for Generative AI Systems](https://mlsanity.com/paper/2503.05336/) (2025)
- [The Problem with Metrics is a Fundamental Problem for AI](https://mlsanity.com/paper/2002.08512/) (2020)
- [BotChat: Evaluating LLMs' Capabilities of Having Multi-Turn Dialogues](https://mlsanity.com/paper/2310.13650/) (2023)
- [Stereotype or Personalization? User Identity Biases Chatbot Recommendations](https://mlsanity.com/paper/2410.05613/) (2024)
- [ChatBench: From Static Benchmarks to Human-AI Evaluation](https://mlsanity.com/paper/2504.07114/) (2025)
- [Exploring Safety-Utility Trade-Offs in Personalized Language Models](https://mlsanity.com/paper/2406.11107/) (2024)
- [The Vulnerability of Language Model Benchmarks: Do They Accurately Reflect True LLM Performance?](https://mlsanity.com/paper/2412.03597/) (2024)
- [Presumed Cultural Identity: How Names Shape LLM Responses](https://mlsanity.com/paper/2502.11995/) (2025)
- [Beyond Prompts: Dynamic Conversational Benchmarking of Large Language Models](https://mlsanity.com/paper/2409.20222/) (2024)
- [First-Person Fairness in Chatbots](https://mlsanity.com/paper/2410.19803/) (2024)
- [A User-Centric Multi-Intent Benchmark for Evaluating Large Language Models](https://mlsanity.com/paper/2404.13940/) (2024)
- [Evaluating the Sensitivity of LLMs to Prior Context](https://mlsanity.com/paper/2506.00069/) (2025)
- [LLM Safety for Children](https://mlsanity.com/paper/2502.12552/) (2025)

## Built on by (most cited first, 8 of 8)

- [One Persona, Many Cues, Different Results: How Sociodemographic Cues Impact LLM Personalization](https://mlsanity.com/paper/2601.18572/) (2026)
- [Different Demographic Cues Yield Inconsistent Conclusions About LLM Personalization and Bias](https://mlsanity.com/paper/2601.18486/) (2026)
- [Do LLMs Favor LLMs? Quantifying Interaction Effects in Peer Review](https://mlsanity.com/paper/2601.20920/) (2026)
- [What Current AI Benchmarks Leave Unmeasured: Modality, Search, Citations, and Implications (for Safety Evaluations)](https://mlsanity.com/paper/2608.06202/) (2026)
- [Protecting patient privacy in clinical foundation models: Technical and legal perspectives](https://mlsanity.com/paper/2608.07705/) (2026)
- [From Controlled to the Wild: Evaluation of Pentesting Agents for the Real-World](https://mlsanity.com/paper/2605.10834/) (2026)
- [Beyond Third-Person Audits: Situated Interaction Auditing for User-Centered LLM Bias Research](https://mlsanity.com/paper/2606.12247/) (2026)
- [API Benchmark Scores Do Not Reliably Transfer to Chatbot Interfaces](https://mlsanity.com/paper/2609.08861/) (2026)

## Abstract (arXiv)

Standard offline evaluations for language models -- a series of independent, state-less inferences made by models -- fail to capture how language models actually behave in practice, where personalization fundamentally alters model behavior. For instance, identical benchmark questions to the same language model can produce markedly different responses when prompted to a state-less system, in one user's chat session, or in a different user's chat session. In this work, we provide empirical evidence showcasing this phenomenon by comparing offline evaluations to field evaluations conducted by having 800 real users of ChatGPT and Gemini pose benchmark and other provided questions to their chat interfaces.