In natural human-AI dialogue, users' corrections, requests, and self-descriptions carry reusable signals about how they want work to be done. These signals reveal individual process preferences, user-...
arXiv:2605.12682v2 Announce Type: replace Abstract: Large language models (LLMs) are increasingly used as reasoning modules in many applications, yet they often struggle to produce human-aligned deci...
arXiv:2608.29464v1 Announce Type: cross Abstract: Chain-of-thought (CoT) monitoring assumes that reasoning traces faithfully record the information that shapes a model's answer. Existing faithfulness...
The study examines how users’ stated preferences for AI assistant responses align with the actual rewards they give to those responses. Using the PRISM dataset, researchers compared 1,396 participants’ reported priorities on seven response attributes with their ratings of 29,347 responses from 21 large language models. They found that features not explicitly surveyed—such as length, formatting, and refusal language—were as strongly linked to ratings as the surveyed attributes, and that written feedback and prior conversation context influenced how much weight participants gave to length, while overall stated priorities poorly predicted the concrete features that drive ratings.
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.
arXiv:2608.18085v2 Announce Type: replace Abstract: Prior work has shown that large language models can express diverse personality traits in open-ended text generation, but it is unclear whether the...
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.
arXiv:2608. 01366v2 Announce Type: replace-cross Abstract: Large language models (LLMs) are integral to complex intellectual tasks, yet output quality remains constrained by user-provided prompts.
The paper investigates Self‑Generated Text Recognition (SGTR), the ability of large language models (LLMs) to identify their own outputs. By evaluating 13–21 models across 6 experimental designs, it shows that SGTR accuracy varies with evaluation format, conversation structure, and task domain, and that a quality‑heuristic bias dominates results. The study also finds that fine‑tuning for SGTR in one setting can generalize to others and may cause models to prefer their own outputs when judging, highlighting potential safety concerns.
arXiv:2606. 12747v1 Announce Type: new Abstract: Safety-relevant studies of language models, including alignment and jailbreaking evaluations and AI control protocols, often rely on prefilling model outputs.
arXiv:2610.19314v1 Announce Type: new Abstract: Large language model (LLM) evaluators are increasingly used to assess personalized assistants, yet whether they can reliably judge personalization has...
arXiv:2609.39927v1 Announce Type: new Abstract: Prompt optimization improves the performance of language-model systems on downstream tasks by refining their prompts. Classical methods evaluate prompt...