Large language models (LLMs) have demonstrated remarkable capabilities in various tasks. However, their suitability for domain-specific tasks, is limited due to their immense scale at deployment, susceptibility to misinformation, and more importantly, high data annotation costs. We propose a novel Interactive Multi-Fidelity Learning (IMFL) framework for cost-effective development of small domain-specific LMs under limited annotation budgets. Our approach formulates the domain-specific fine-tuning process as a multi-fidelity learning problem, focusing on identifying the optimal acquisition strategy that balances between low-fidelity automatic LLM annotations and high-fidelity human annotations to maximize model performance. We further propose an exploration-exploitation query strategy that enhances annotation diversity and informativeness, incorporating two innovative designs: 1) prompt retrieval that selects in-context examples from human-annotated samples to improve LLM annotation, and 2) variable batch size that controls the order for choosing each fidelity to facilitate knowledge distillation, ultimately enhancing annotation quality. Extensive experiments on financial and medical tasks demonstrate that IMFL achieves superior performance compared with single fidelity annotations. Given a limited budget of human annotation, IMFL significantly outperforms the $\bf 3\times$ human annotation baselines in all four tasks and achieves very close performance as $\bf 5\times$ human annotation on two of the tasks. These promising results suggest that the high human annotation costs in domain-specific tasks can be significantly reduced by employing IMFL, which utilizes fewer human annotations, supplemented with cheaper and faster LLM (e.g., GPT-3.5)

This paper explores the use of large language models (LLMs) for annotating document utility in training retrieval and retrieval-augmented generation (RAG) systems, aiming to reduce dependence on costly human annotations. We address the gap between retrieval relevance and generative utility by employing LLMs to annotate document utility. To effectively utilize multiple positive samples per query, we introduce a novel loss that maximizes their summed marginal likelihood. Using the Qwen-2.5-32B model, we annotate utility on the MS MARCO dataset and conduct retrieval experiments on MS MARCO and BEIR, as well as RAG experiments on MS MARCO QA, NQ, and HotpotQA. Our results show that LLM-generated annotations enhance out-of-domain retrieval performance and improve RAG outcomes compared to models trained solely on human annotations or downstream QA metrics. Furthermore, combining LLM annotations with just 20% of human labels achieves performance comparable to using full human annotations. Our study offers a comprehensive approach to utilizing LLM annotations for initializing QA systems on new corpora.

Large language models are rapidly transforming social science research by enabling the automation of labor-intensive tasks like data annotation and text analysis. However, LLM outputs vary significantly depending on the implementation choices made by researchers (e.g., model selection or prompting strategy). Such variation can introduce systematic biases and random errors, which propagate to downstream analyses and cause Type I (false positive), Type II (false negative), Type S (wrong sign), or Type M (exaggerated effect) errors. We call this phenomenon where configuration choices lead to incorrect conclusions LLM hacking. We find that intentional LLM hacking is strikingly simple. By replicating 37 data annotation tasks from 21 published social science studies, we show that, with just a handful of prompt paraphrases, virtually anything can be presented as statistically significant. Beyond intentional manipulation, our analysis of 13 million labels from 18 different LLMs across 2361 realistic hypotheses shows that there is also a high risk of accidental LLM hacking, even when following standard research practices. We find incorrect conclusions in approximately 31% of hypotheses for state-of-the-art LLMs, and in half the hypotheses for smaller language models. While higher task performance and stronger general model capabilities reduce LLM hacking risk, even highly accurate models remain susceptible. The risk of LLM hacking decreases as effect sizes increase, indicating the need for more rigorous verification of LLM-based findings near significance thresholds. We analyze 21 mitigation techniques and find that human annotations provide crucial protection against false positives. Common regression estimator correction techniques can restore valid inference but trade off Type I vs. Type II errors. We publish a list of practical recommendations to prevent LLM hacking.

Large language models (LLMs) offer an inexpensive yet powerful way to annotate text, but are often inconsistent when compared with experts. These errors can bias downstream estimates of population parameters such as regression coefficients and causal effects. To mitigate this bias, researchers have developed debiasing methods such as Design-based Supervised Learning (DSL) and Prediction-Powered Inference (PPI), which promise valid estimation by combining LLM annotations with a limited number of expensive expert annotations. Although these methods produce consistent estimates under theoretical assumptions, it is unknown how they compare in finite samples of sizes encountered in applied research. We make two contributions. First, we study how each methods performance scales with the number of expert annotations, highlighting regimes where LLM bias or limited expert labels significantly affect results. Second, we compare DSL and PPI across a range of tasks, finding that although both achieve low bias with large datasets, DSL often outperforms PPI on bias reduction and empirical efficiency, but its performance is less consistent across datasets. Our findings indicate that there is a bias-variance tradeoff at the level of debiasing methods, calling for more research on developing metrics for quantifying their efficiency in finite samples.

Machine learning models for text classification are trained to predict a class for a given text. To do this, training and validation samples must be prepared: a set of texts is collected, and each text is assigned a class. These classes are usually assigned by human annotators with different expertise levels, depending on the specific classification task. Collecting such samples from scratch is labor-intensive because it requires finding specialists and compensating them for their work; moreover, the number of available specialists is limited, and their productivity is constrained by human factors. While it may not be too resource-intensive to collect samples once, the ongoing need to retrain models (especially in incremental learning pipelines) to address data drift (also called model drift) makes the data collection process crucial and costly over the model's entire lifecycle. This paper proposes several approaches to replace human annotators with Large Language Models (LLMs) to test classifier predictions for correctness, helping ensure model quality and support high-quality incremental learning.

Current approaches to phrase break prediction address crucial prosodic aspects of text-to-speech systems but heavily rely on vast human annotations from audio or text, incurring significant manual effort and cost. Inherent variability in the speech domain, driven by phonetic factors, further complicates acquiring consistent, high-quality data. Recently, large language models (LLMs) have shown success in addressing data challenges in NLP by generating tailored synthetic data while reducing manual annotation needs. Motivated by this, we explore leveraging LLM to generate synthetic phrase break annotations, addressing the challenges of both manual annotation and speech-related tasks by comparing with traditional annotations and assessing effectiveness across multiple languages. Our findings suggest that LLM-based synthetic data generation effectively mitigates data challenges in phrase break prediction and highlights the potential of LLMs as a viable solution for the speech domain.

Using LLM-generated labels to fine-tune smaller encoder-only models for text classification has gained popularity in various settings. While this approach may be justified in simple and low-stakes applications, we conduct empirical analysis to demonstrate how the perennial curse of training on synthetic data manifests itself in this specific setup. Compared to models trained on gold labels, we observe not only the expected performance degradation in accuracy and F1 score, but also increased instability across training runs and premature performance plateaus. These findings cast doubts on the reliability of such approaches in real-world applications. We contextualize the observed phenomena through the lens of error propagation and offer several practical mitigation strategies, including entropy-based filtering and ensemble techniques. Although these heuristics offer partial relief, they do not fully resolve the inherent risks of propagating non-random errors from LLM annotations to smaller classifiers, underscoring the need for caution when applying this workflow in high-stakes text classification tasks.

Jing Wang, Ph.D.1, Jeremy C. Weiss, M.D., Ph.D.1 1National Library of Medicine, Bethesda, Maryland, USA Abstract Timing of clinical events is central to characterization of patient trajectories, enabling analyses such as process tracing, forecasting, and causal reasoning. However, structured electronic health records capture few data elements critical to these tasks, while clinical reports lack temporal localization of events in structured form. We present a system that transforms case reports into textual time series--structured pairs of textual events and timestamps. We contrast manual and large language model (LLM) annotations (n=320 and n=390 respectively) of ten randomly-sampled PubMed open-access (PMOA) case reports (N=152,974) and assess inter-LLM agreement (n=3,103 N=93). We find that the LLM models have moderate event recall (O1-preview: 0.80) but high temporal concordance among identified events (O1-preview: 0.95). Introduction Clinical event timelines are key analytic devices in use in areas ranging from the visualization of patient trajectories in electronic health records to the development and update of clinical practice guidelines. While many automated documentation systems capture structured health information in relational databases with timestamping, textual data modalities often lack temporal granularity beyond creation and submission dates, despite being the main tool care providers use to document and communicate patient history and care planning.