Language Model Generation Evaluation

The transition of Large Language Models (LLMs) from exploratory tools to active "silicon subjects" in social science lacks extensive validation of operational validity. This study introduces Conditioned Comment Prediction (CCP), a task in which a model predicts how a user would comment on a given stimulus by comparing generated outputs with authentic digital traces. This framework enables a rigorous evaluation of current LLM capabilities with respect to the simulation of social media user behavior. We evaluated open-weight 8B models (Llama3.1, Qwen3, Ministral) in English, German, and Luxembourgish language scenarios. By systematically comparing prompting strategies (explicit vs. implicit) and the impact of Supervised Fine-Tuning (SFT), we identify a critical form vs. content decoupling in low-resource settings: while SFT aligns the surface structure of the text output (length and syntax), it degrades semantic grounding. Furthermore, we demonstrate that explicit conditioning (generated biographies) becomes redundant under fine-tuning, as models successfully perform latent inference directly from behavioral histories. Our findings challenge current "naive prompting" paradigms and offer operational guidelines prioritizing authentic behavioral traces over descriptive personas for high-fidelity simulation.

Human evaluations play a central role in training and assessing AI models, yet these data are rarely treated as measurements subject to systematic error. This paper integrates psychometric rater models into the AI pipeline to improve the reliability and validity of conclusions drawn from human judgments. The paper reviews common rater effects, severity and centrality, that distort observed ratings, and demonstrates how item response theory rater models, particularly the multi-faceted Rasch model, can separate true output quality from rater behavior. Using the OpenAI summarization dataset as an empirical example, we show how adjusting for rater severity produces corrected estimates of summary quality and provides diagnostic insight into rater performance. Incorporating psychometric modeling into human-in-the-loop evaluation offers more principled and transparent use of human data, enabling developers to make decisions based on adjusted scores rather than raw, error-prone ratings. This perspective highlights a path toward more robust, interpretable, and construct-aligned practices for AI development and evaluation.

In AI, most evaluations of natural language understanding tasks are conducted in standardized dialects such as Standard American English (SAE). In this work, we investigate how accurately large language models (LLMs) represent African American Vernacular English (AAVE). We analyze three LLMs to compare their usage of AAVE to the usage of humans who natively speak AAVE. We first analyzed interviews from the Corpus of Regional African American Language and TwitterAAE to identify the typical contexts where people use AAVE grammatical features such as ain't. We then prompted the LLMs to produce text in AAVE and compared the model-generated text to human usage patterns. We find that, in many cases, there are substantial differences between AAVE usage in LLMs and humans: LLMs usually underuse and misuse grammatical features characteristic of AAVE. Furthermore, through sentiment analysis and manual inspection, we found that the models replicated stereotypes about African Americans. These results highlight the need for more diversity in training data and the incorporation of fairness methods to mitigate the perpetuation of stereotypes.

Structured Outputs from current LLMs exhibit sporadic errors, hindering enterprise AI deployment. We present CONSTRUCT, a real-time uncertainty estimator that scores the trustworthiness of LLM Structured Outputs. Lower-scoring outputs are more likely to contain errors, enabling automatic prioritization of limited human review bandwidth. CONSTRUCT additionally scores the trustworthiness of each field within a Structured Output, helping reviewers quickly identify which parts of the output are incorrect. Our method is suitable for any LLM (including black-box LLM APIs without logprobs), does not require labeled training data or custom model deployment, and supports complex Structured Outputs with heterogeneous fields and nested JSON schemas. We also introduce one of the first public LLM Structured Output benchmarks with reliable ground-truth values. Over this four-dataset benchmark, CONSTRUCT detects errors in outputs from various LLMs (including Gemini 3 and GPT-5) with significantly higher precision/recall than existing techniques.

Large language models (LLMs) demonstrate the capacity to reconstruct and trace learned content from their training data under specific elicitation conditions, yet this capability does not manifest in standard generation contexts. This empirical observational study examines the expression of non-causal, non-implementable solution types across 300 prompt-response generations spanning narrative and problem-solving task contexts. Drawing on recent findings regarding memorization contiguity and alignment-induced discourse priors, we document a systematic dissociation between learned capability and expressed output. Across three distinct LLMs, ten task scenarios, and both creative narrative and practical advisory contexts, we documented zero instances of non-causal solution frames in generated outputs (0%, 95% CI: [0%, 1.2%]), despite verified reconstruction capability under conditional extraction. These findings challenge the prevailing assumption that training data presence directly predicts output probability, demonstrating instead that task-conditioned generation policies can comprehensively suppress learned content across diverse contexts. The results offer implications for understanding generation dynamics, output distribution control, and the behavioral boundaries of contemporary LLMs.

The increasing use of Large Language Models (LLMs) as proxies for human participants in social science research presents a promising, yet methodologically risky, paradigm shift. While LLMs offer scalability and cost-efficiency, their "naive" application, where they are prompted to generate content without explicit behavioral constraints, introduces significant linguistic discrepancies that challenge the validity of research findings. This paper addresses these limitations by introducing a novel, history-conditioned reply prediction task on authentic X (formerly Twitter) data, to create a dataset designed to evaluate the linguistic output of LLMs against human-generated content. We analyze these discrepancies using stylistic and content-based metrics, providing a quantitative framework for researchers to assess the quality and authenticity of synthetic data. Our findings highlight the need for more sophisticated prompting techniques and specialized datasets to ensure that LLM-generated content accurately reflects the complex linguistic patterns of human communication, thereby improving the validity of computational social science studies.

Robust and comprehensive evaluation of large language models (LLMs) is essential for identifying effective LLM system configurations and mitigating risks associated with deploying LLMs in sensitive domains. However, traditional statistical metrics are poorly suited to open-ended generation tasks, leading to growing reliance on LLM-based evaluation methods. These methods, while often more flexible, introduce additional complexity: they depend on carefully chosen models, prompts, parameters, and evaluation strategies, making the evaluation process prone to misconfiguration and bias. In this work, we present EvalSense, a flexible, extensible framework for constructing domain-specific evaluation suites for LLMs. EvalSense provides out-of-the-box support for a broad range of model providers and evaluation strategies, and assists users in selecting and deploying suitable evaluation methods for their specific use-cases. This is achieved through two unique components: (1) an interactive guide aiding users in evaluation method selection and (2) automated meta-evaluation tools that assess the reliability of different evaluation approaches using perturbed data. We demonstrate the effectiveness of EvalSense in a case study involving the generation of clinical notes from unstructured doctor-patient dialogues, using a popular open dataset. All code, documentation, and assets associated with EvalSense are open-source and publicly available at https://github.com/nhsengland/evalsense.

We present an end-to-end self-hosted (API-free) pipeline, where API-free means that lecture content is not sent to any external LLM service, that converts lecture PDFs into multiple-choice questions (MCQs) using a local LLM plus deterministic quality control (QC). The pipeline is designed for black-box minimization: LLMs may assist drafting, but the final released artifacts are plain-text question banks with an explicit QC trace and without any need to call an LLM at deployment time. We run a seed sweep on three short "dummy lectures" (information theory, thermodynamics, and statistical mechanics), collecting 15 runs x 8 questions = 120 accepted candidates (122 attempts total under bounded retries). All 120 accepted candidates satisfy hard QC checks (JSON schema conformance, a single marked correct option, and numeric/constant equivalence tests); however, the warning layer flags 8/120 items (spanning 8 runs) that expose residual quality risks such as duplicated distractors or missing rounding instructions. We report a warning taxonomy with concrete before->after fixes, and we release the final 24-question set (three lectures x 8 questions) as JSONL/CSV for Google Forms import (e.g., via Apps Script or API tooling) included as ancillary files under anc/. Finally, we position the work through the AI to Learn (AI2L) rubric lens and argue that self-hosted MCQ generation with explicit QC supports privacy, accountability, and Green AI in educational workflows.

Evaluating the reasoning capabilities of Large Language Models is increasingly challenging as models improve. Human curation of hard questions is highly expensive, especially in recent benchmarks using PhD-level domain knowledge to challenge the most capable models. Even then, there is always a concern about whether these questions test genuine reasoning or if similar problems have been seen during training. Here, we take inspiration from 16th-century mathematical duels to design The Token Games (TTG): an evaluation framework where models challenge each other by creating their own puzzles. We leverage the format of Programming Puzzles - given a function that returns a boolean, find inputs that make it return True - to flexibly represent problems and enable verifying solutions. Using results from pairwise duels, we then compute Elo ratings, allowing us to compare models relative to each other. We evaluate 10 frontier models on TTG, and closely match the ranking from existing benchmarks such as Humanity's Last Exam, spending less than $200 USD and without involving any human effort in creating puzzles. We also find that creating good puzzles is still a highly challenging task for current models. Overall, our work suggests new paradigms for evaluating reasoning that avoid saturation by design, and that allow testing models for other skills like creativity and task creation alongside problem solving.

Meenzerisch, the dialect spoken in the German city of Mainz, is also the traditional language of the Mainz carnival, a yearly celebration well known throughout Germany. However, Meenzerisch is on the verge of dying out-a fate it shares with many other German dialects. Natural language processing (NLP) has the potential to help with the preservation and revival efforts of languages and dialects. However, so far no NLP research has looked at Meenzerisch. This work presents the first research in the field of NLP that is explicitly focused on the dialect of Mainz. We introduce a digital dictionary-an NLP-ready dataset derived from an existing resource (Schramm, 1966)-to support researchers in modeling and benchmarking the language. It contains 2,351 words in the dialect paired with their meanings described in Standard German. We then use this dataset to answer the following research questions: (1) Can state-of-the-art large language models (LLMs) generate definitions for dialect words? (2) Can LLMs generate words in Meenzerisch, given their definitions? Our experiments show that LLMs can do neither: the best model for definitions reaches only 6.27% accuracy and the best word generation model's accuracy is 1.51%. We then conduct two additional experiments in order to see if accuracy is improved by few-shot learning and by extracting rules from the training set, which are then passed to the LLM. While those approaches are able to improve the results, accuracy remains below 10%. This highlights that additional resources and an intensification of research efforts focused on German dialects are desperately needed.

Large language models (LLMs) are increasingly applied as automatic evaluators for natural language generation assessment often using pairwise comparative judgements. Existing approaches typically rely on single judges or aggregate multiple judges assuming equal reliability. In practice, LLM judges vary substantially in performance across tasks and evaluation aspects, and their judgment probabilities may be biased and inconsistent. Furthermore, human-labelled supervision for judge calibration may be unavailable. We first empirically demonstrate that inconsistencies in LLM comparison probabilities exist and show that it limits the effectiveness of direct probability-based ranking. To address this, we study the LLM-asa-jury setting and propose BT-sigma, a judge-aware extension of the Bradley-Terry model that introduces a discriminator parameter for each judge to jointly infer item rankings and judge reliability from pairwise comparisons alone. Experiments on benchmark NLG evaluation datasets show that BT-sigma consistently outperforms averaging-based aggregation methods, and that the learned discriminators strongly correlate with independent measures of the cycle consistency of LLM judgments. Further analysis reveals that BT-sigma can be interpreted as an unsupervised calibration mechanism that improves aggregation by modelling judge reliability.

Can a machine write good poetry? Any positive answer raises fundamental questions about the nature and value of art. We report a seven-month poetry workshop in which a large language model was shaped into a digital poet through iterative in-context expert feedback, without retraining. Across sessions, the model developed a distinctive style and a coherent corpus, supported by quantitative and qualitative analyses, and it produced a pen name and author image. In a blinded authorship test with 50 humanities students and graduates (three AI poems and three poems by well-known poets each), judgments were at chance: human poems were labeled human 54% of the time and AI poems 52%, with 95% confidence intervals including 50%. After the workshop, a commercial publisher released a poetry collection authored by the model. These results show that workshop-style prompting can support long-horizon creative shaping and renew debates on creativity and authorship.

We argue that uncertainty is a key and understudied limitation of LLMs' performance in creative writing, which is often characterized as trite and cliché-ridden. Literary theory identifies uncertainty as a necessary condition for creative expression, while current alignment strategies steer models away from uncertain outputs to ensure factuality and reduce hallucination. We formalize this tension by quantifying the ``uncertainty gap'' between human-authored stories and model-generated continuations. Through a controlled information-theoretic analysis of 28 LLMs on high-quality storytelling datasets, we demonstrate that human writing consistently exhibits significantly higher uncertainty than model outputs. We find that instruction-tuned and reasoning models exacerbate this trend compared to their base counterparts; furthermore, the gap is more pronounced in creative writing than in functional domains, and shows a consistent correlation with writing quality. Achieving human-level creativity requires new uncertainty-aware alignment paradigms that can distinguish between destructive hallucinations and the constructive ambiguity required for literary richness.

Large language models (LLMs) are increasingly used as agents to solve complex tasks such as question answering (QA), scientific debate, and software development. A standard evaluation procedure aggregates multiple responses from LLM agents into a single final answer, often via majority voting, and compares it against reference answers. However, this process can obscure the quality and distributional characteristics of the original responses. In this paper, we propose a novel evaluation framework based on the empirical cumulative distribution function (ECDF) of cosine similarities between generated responses and reference answers. This enables a more nuanced assessment of response quality beyond exact match metrics. To analyze the response distributions across different agent configurations, we further introduce a clustering method for ECDFs using their distances and the $k$-medoids algorithm. Our experiments on a QA dataset demonstrate that ECDFs can distinguish between agent settings with similar final accuracies but different quality distributions. The clustering analysis also reveals interpretable group structures in the responses, offering insights into the impact of temperature, persona, and question topics.

Evaluating the quality of automatically generated text often relies on LLM-as-a-judge (LLM-judge) methods. While effective, these approaches are computationally expensive and require post-processing. To address these limitations, we build upon ParaPLUIE, a perplexity-based LLM-judge metric that estimates confidence over ``Yes/No'' answers without generating text. We introduce *-PLUIE, task specific prompting variants of ParaPLUIE and evaluate their alignment with human judgement. Our experiments show that personalised *-PLUIE achieves stronger correlations with human ratings while maintaining low computational cost.

Producing output that conforms to a specified JSON schema underlies tool use, structured extraction, and knowledge base construction in modern large language models. Despite this centrality, public datasets for the task remain small, synthetic, or text-only, and rarely pair real page content with the prompts and schemas used in practice. We introduce ScrapeGraphAI-100k, 93,695 schema-constrained extraction events collected via opt-in ScrapeGraphAI telemetry in Q2--Q3 2025, deduplicated and balanced by schema from 9M raw events. The corpus spans 18 000+ unique schemas across 15 named languages plus a long-tail Other category, with English and Traditional Chinese covering 88% of detected content, each instance pairs Markdown-converted page content with a prompt, schema, LLM response, and per-example jsonschema-rs structural conformance labels (semantic correctness is out of scope, and raw HTML is deferred beyond v1.0). We characterize structural diversity across the corpus and identify sharp failure thresholds as schema complexity grows. As a case study, a 1.7B student fine-tuned on this data closely tracks the output distribution of its GPT-5-nano teacher, though it still trails a 30B-A3B reference (3.3B active parameters) on schema compliance. We offer this distillation result as preliminary evidence that grounding schema-constrained generation in real practitioner workloads at scale enables training and benchmarking that prior synthetic or text-only corpora could not support.

Prior work has explored multi-turn interaction and feedback for LLM writing, but evaluations still largely center on prompts and localized feedback, leaving persistent public reception in online communities underexamined. We test whether broadcast community discussion improves stand-up comedy writing in a controlled multi-agent sandbox: in the discussion condition, critic and audience threads are recorded, filtered, stored as social memory, and later retrieved to condition subsequent generations, whereas the baseline omits discussion. Across 50 rounds (250 paired monologues) judged by five expert annotators using A/B preference and a 15-item rubric, discussion wins 75.6% of instances and improves Craft/Clarity (Δ = 0.440) and Social Response (Δ = 0.422), with occasional increases in aggressive humor.

We consider the task of end-to-end spreadsheet generation, where language models produce spreadsheet artifacts to satisfy users' explicit and implicit constraints, specified in natural language. We introduce SpreadsheetArena, a platform for evaluating models' performance on the task via blind pairwise preference votes of LLM-generated spreadsheet workbooks. As with other complex, open-ended tasks, relevant evaluation criteria can vary greatly across use cases, often in ways that are difficult to formalize. Compared to general dialogue or text generation settings, spreadsheet generation presents unique challenges and opportunities: the task output structure is well-defined and multi-dimensional, and there are often complex interactivity and layout considerations. We observe that stylistic, structural, and functional features of preferred spreadsheets vary meaningfully across prompts. Expert evaluations of spreadsheets for finance prompts suggest that even highly ranked models do not reliably produce spreadsheets aligned with domain-specific best practices. We host a live arena and release a dataset of prompts, generated spreadsheets, and preference votes, which we hope will facilitate further study of tasks operating over spreadsheets as a challenging and interesting class of complex, open-ended tasks for LLMs.

LLM-as-a-judge ensembles are the standard paradigm for scalable evaluation, but their aggregation mechanisms suffer from a fundamental flaw: they implicitly assume that judges provide independent estimates of true quality. However, in practice, LLM judges exhibit correlated errors caused by shared latent confounders -- such as verbosity, stylistic preferences, or training artifacts -- causing standard aggregation rules like majority vote or averaging to provide little gain or even amplify systematic mistakes. To address this, we introduce CARE, a confounder-aware aggregation framework that explicitly models LLM judge scores as arising from both a latent true-quality signal and shared confounding factors. Rather than heuristically re-weighting judges, CARE separates quality from confounders without access to ground-truth labels. We provide theoretical guarantees for identifiability and finite-sample recovery under shared confounders, and we quantify the systematic bias incurred when aggregation models omit confounding latent factors. Across 12 public benchmarks spanning continuous scoring, binary classification, and pairwise preference settings, CARE improves aggregation accuracy, reducing error by up to 26.8%. Code is released in https://github.com/SprocketLab/CARE.

Recent research has shown that large language models (LLMs) favor their own outputs when acting as judges, undermining the integrity of automated post-training and evaluation workflows. However, it is difficult to disentangle which behaviors are explained by narcissism versus experimental confounds. Specifically, LLM evaluators may deliver self-preferring verdicts when comparing responses to questions they fail on; these verdicts may not depend on the identity of the author, but on evaluator quality. We correct this by directly comparing the judge's voting distribution in cases where it evaluates itself versus another model. This evaluator quality baseline reveals that only 51% of examples in previous findings retain statistical significance against this null hypothesis, covering 89.6% of total self-preference probability mass. Finally, we compare the entropy of voting distributions, suggesting uncertainty-driven overlap, and show that our procedure enables more careful documentation against the backdrop of judge-bias research.

As Natural Language Generation (NLG) dominates modern NLP, scalable evaluation remains a critical bottleneck. Consequently, LLM-as-a-judge (LaaJ) adoption has accelerated rapidly, appearing in more papers than human evaluation in 2025. This pivotal shift motivates a critical analysis of current evaluation practices. Overcoming the limits of rigid keyword filtering and manual review, we employ a multi-LLM information extraction pipeline to gather structured metadata from 14,171 papers across four major NLP conferences (2020-2025). Analyzing 3,334 filtered NLG papers, we identify three systemic challenges. (1) Metric inertia: despite the shift toward open-ended generation, legacy lexical metrics (BLEU, ROUGE) persist as primary indicators, typically used alongside rather than replaced by semantic alternatives. (2) Metric-criteria mapping problem: our paper-level co-occurrence data reveals that general-purpose automatic metrics are applied as broad proxies for quality, without specifying which dimension of text generation they are intended to evaluate. (3) Validation gap: LaaJ has grown rapidly without commensurate human validation (fewer than 8% of papers). Crucially, while LaaJ correlates with aggregate quality, alignment collapses on fine-grained criteria like fluency. To address these gaps, we distill our findings into a minimal Evaluation Checklist to guide metric selection, construct validity, and LaaJ deployment.

As frontier AI systems are pretrained on web-scale data, test set contamination has become a critical concern for accurately assessing their capabilities. While research has thoroughly investigated the impact of test set contamination on discriminative evaluations like multiple-choice question-answering, comparatively little research has studied the impact of test set contamination on generative evaluations. In this work, we quantitatively assess the effect of test set contamination on generative evaluations through the language model lifecycle. We pretrain language models on mixtures of web data and the MATH benchmark, sweeping model sizes and number of test set replicas contaminating the pretraining corpus; performance improves with contamination and model size. Using scaling laws, we make a surprising discovery: including even a single test set replica enables models to achieve lower loss than the irreducible error of training on the uncontaminated corpus. We then study further training: overtraining with fresh data reduces the effects of contamination, whereas supervised finetuning on the training set can either increase or decrease performance on test data, depending on the amount of pretraining contamination. Finally, at inference, we identify factors that modulate memorization: high sampling temperatures mitigate contamination effects, and longer solutions are exponentially more difficult to memorize than shorter ones, presenting a contrast with discriminative evaluations, where solutions are only a few tokens in length. By characterizing how generation and memorization interact, we highlight a new layer of complexity for trustworthy evaluation of AI systems.

It is increasingly important to evaluate the characteristics of systems based on large language models (LLMs). Evaluations in this context often rely on a curated benchmark set of input prompts provided to the LLM, where the output for each prompt may be assigned a binary or ordinal score and the aggregation of scores across prompts is then used as a summary evaluation. In this paper, we develop a Bayesian approach for quantifying the uncertainty that arises in such evaluation metrics as a result of the stochasticity of the LLM-based systems; the same prompt may exhibit different outcomes on repeated runs. Our framework leads naturally to a sequential evaluation, in which we leverage the Bayesian model to preferentially select which prompts in the benchmark to use next, enabling more cost-effective LLM evaluations. We demonstrate this approach through four case studies: pairwise LLM preferences in interactive dialogue (MT-Bench), refusal rates on adversarial inputs (JailbreakBench), embodied agentic reasoning (ALFWorld), and agentic web-search Q&A (GAIA). Our results illustrate how a Bayesian approach can provide insightful uncertainty quantification about the behavior of blackbox LLM-based systems in a manner that is not accessible via deterministic or single-run evaluations.

Language models (LMs) often struggle to generate diverse, human-like creative content, raising concerns about the long-term homogenization of human thought through repeated exposure to similar outputs. Yet scalable methods for evaluating LM output diversity remain limited, especially beyond narrow tasks such as random number or name generation, or beyond repeated sampling from a single model. We introduce Infinity-Chat, a large-scale dataset of 26K diverse, real-world, open-ended user queries that admit a wide range of plausible answers with no single ground truth. We introduce the first comprehensive taxonomy for characterizing the full spectrum of open-ended prompts posed to LMs, comprising 6 top-level categories (e.g., brainstorm & ideation) that further breaks down to 17 subcategories. Using Infinity-Chat, we present a large-scale study of mode collapse in LMs, revealing a pronounced Artificial Hivemind effect in open-ended generation of LMs, characterized by (1) intra-model repetition, where a single model consistently generates similar responses, and more so (2) inter-model homogeneity, where different models produce strikingly similar outputs. Infinity-Chat also includes 31,250 human annotations, across absolute ratings and pairwise preferences, with 25 independent human annotations per example. This enables studying collective and individual-specific human preferences in response to open-ended queries. Our findings show that LMs, reward models, and LM judges are less well calibrated to human ratings on model generations that elicit differing idiosyncratic annotator preferences, despite maintaining comparable overall quality. Overall, INFINITY-CHAT presents the first large-scale resource for systematically studying real-world open-ended queries to LMs, revealing critical insights to guide future research for mitigating long-term AI safety risks posed by the Artificial Hivemind.

Standard language model evaluation assigns scores to single predicted answers, rewarding high-confidence responses regardless of how residual probability mass is distributed over alternative options. This creates a systematic pressure toward overconfident guessing: under accuracy-based schemes, a model maximises its expected score by always committing to an answer rather than abstaining, even when its uncertainty is high. While penalty-based approaches partially address this by raising the confidence threshold for strategic guessing, they still treat all sub-threshold responses identically, ignoring a fundamental distinction in how models can express uncertainty - for example between hedging toward incorrect answers versus hedging toward "I don't know" responses. This paper introduces a novel evaluation metric to solve this problem of not considering a model's entire probability distribution over answer choices. The metric naturally distinguishes between harmful overconfidence in wrong answers and uncertainty expressed through abstention, providing scores in an interpretable default range. Through theoretical analysis and illustrative examples, the metric is shown to offer a more nuanced and aligned evaluation paradigm that incentivises models to express genuine uncertainty rather than guessing. Adapting 12 existing evaluation benchmarks to the metric's variants and measuring performance on six language models shows that for half of the tested benchmarks scores are negative across all tested models, indicating significant tendencies towards hallucination.

As LLMs rapidly saturate existing benchmarks, automated benchmark creation using LLMs (LLM as a benchmark) where a model generates test inputs (LLM as a testset) and evaluates outputs (LLM as an evaluator) has gained traction as a cheap alternative to human curation. We show that this paradigm has a fundamental problem: LLM-generated benchmarks systematically favor the model that created them. Using machine translation as our primary testbed, we find that self bias arises from two compounding sources, LLM as a testset and LLM as an evaluator, and their combination amplifies the effect. Crucially, even when test data is generated with explicit diversity controls, each modelś implicit stylistic tendencies produce homogeneous, model-specific outputs that inflate its own scores. Increasing source text diversity, using our proposed diversity metric, partially mitigates this bias. Self bias is strong enough to cause each model to rank itself first, overriding the peer consensus ordering. We confirm that the phenomenon extends to open-ended generation on the Chatbot Arena task.

AI "slop" is an increasingly popular term used to describe low-quality AI-generated text, but there is currently no agreed upon definition of this term nor a means to measure its occurrence. In this work, we develop a taxonomy of "slop" through interviews with experts in NLP, writing, and philosophy, and propose a set of interpretable dimensions for its assessment in text. Through span-level annotation, we find that binary "slop" judgments are (somewhat) subjective, but such determinations nonetheless correlate with latent dimensions such as coherence and relevance. Our framework can be used to examine AI-generated text in both detection and binary preference tasks, potentially offering new insights into the linguistic and stylistic factors that contribute to quality judgments. We highlight that fully automated and scalable methods remain an open challenge.

A controversial test for Large Language Models concerns the ability to discern possible from impossible language. While some evidence attests to the models sensitivity to what crosses the limits of grammatically impossible language, this evidence has been contested on the grounds of the soundness of the testing material. We use model-internal representations as a direct means of investigating the way Large Language Models represent the grammatical-ungrammatical distinction. In a novel benchmark, we elicit sum log probabilities from 8 models and compute minimal-pair surprisal differences, juxtaposing probabilities assigned to grammatical sentences to probabilities assigned to (i) lower frequency grammatical sentences, (ii) ungrammatical sentences, (iii) semantically odd sentences, and (iv) pragmatically odd sentences. The prediction is that if string-probabilities can function as proxies for the limits of grammar, the ungrammatical condition will stand out among the conditions that involve linguistic violations, showing a spike in the surprisal rates. Our results do not reveal a unique surprisal signature for ungrammatical prompts: across models, we observe comparable surprisal for the pragmatically odd condition, and the semantically odd condition consistently shows even higher surprisal. We thus demonstrate that probabilities do not constitute reliable proxies for grammaticality, showing that probability differences alone cannot be used to infer model-internal representations of syntactic knowledge. Consequently, claims about models being able to distinguish possible from impossible language need verification through a different methodology.

As researchers increasingly adopt LLMs as writing assistants, generating high-quality research paper introductions remains both challenging and essential. We introduce Scientific Introduction Generation (SciIG), a task that evaluates LLMs' ability to produce coherent introductions from titles, abstracts, and related works. Curating new datasets from NAACL 2025 and ICLR 2025 papers, we assess five state-of-the-art models, including both open-source (DeepSeek-v3, Gemma-3-12B, LLaMA 4-Maverick, MistralAI Small 3.1) and closed-source GPT-4o systems, across multiple dimensions: lexical overlap, semantic similarity, content coverage, faithfulness, consistency, citation correctness, and narrative quality. Our comprehensive framework combines automated metrics with LLM-as-a-judge evaluations. Results demonstrate LLaMA-4 Maverick's superior performance on most metrics, particularly in semantic similarity and faithfulness. Moreover, three-shot prompting consistently outperforms fewer-shot approaches. These findings provide practical insights into developing effective research writing assistants and set realistic expectations for LLM-assisted academic writing. To foster re- producibility and future research, we publicly release all code and datasets.

With the surge of large language models (LLMs) and their ability to produce customized output, style-personalized text generation--"write like me"--has become a rapidly growing area of interest. However, style personalization is highly specific, relative to every user, and depends strongly on the pragmatic context, which makes it uniquely challenging. Although prior research has introduced benchmarks and metrics for this area, they tend to be non-standardized and have known limitations (e.g., poor correlation with human subjects). LLMs have been found to not capture author-specific style well, it follows that the metrics themselves must be scrutinized carefully. In this work we critically examine the effectiveness of the most common metrics used in the field, such as BLEU, embeddings, and LLMs-as-judges. We evaluate these metrics using our proposed style discrimination benchmark, which spans eight diverse writing tasks across three evaluation settings: domain discrimination, authorship attribution, and LLM-generated personalized vs non-personalized discrimination. We find strong evidence that employing ensembles of diverse evaluation metrics consistently outperforms single-evaluator methods, and conclude by providing guidance on how to reliably assess style-personalized text generation.