Ensuring that Large Language Models (LLMs) generate text representative of diverse sub-populations is essential, particularly when key concepts related to under-represented groups are scarce in the training data. We address this challenge with a novel clustering-based active learning framework, enhanced with knowledge distillation. The proposed framework transforms the intermediate outputs of the learner model, enabling effective active learning for generative tasks for the first time. Integration of clustering and knowledge distillation yields more representative models without prior knowledge of underlying data distribution and overbearing human efforts. We validate our approach in practice through case studies in counter-narration and style transfer. We construct two new datasets in tandem with model training, showing a performance improvement of 2%–10% over baseline models. Our results also show more consistent performance across various data subgroups and increased lexical diversity, underscoring our model’s resilience to skewness in available data. Further, our results show that the data acquired via our approach improves the performance of secondary models not involved in the learning loop, showcasing practical utility of the framework.

General-purpose automatic speech recognition (ASR) systems do not always perform well in goal-oriented dialogue. Existing ASR correction methods rely on prior user data or named entities. We extend correction to tasks that have no prior user data and exhibit linguistic flexibility such as lexical and syntactic variations. We propose a novel context augmentation with a large language model and a ranking strategy that incorporates contextual information from the dialogue states of a goal-oriented conversational AI and its tasks. Our method ranks (1) n-best ASR hypotheses by their lexical and semantic similarity with context and (2) context by phonetic correspondence with ASR hypotheses. Evaluated in home improvement and cooking domains with real-world users, our method improves recall and F1 of correction by 34% and 16%, respectively, while maintaining precision and false positive rate. Users rated .8-1 point (out of 5) higher when our correction method worked properly, with no decrease due to false positives.

Large vision language models (VLMs) increasingly claim reasoning skills, yet current benchmarks evaluate them in single-turn or question answering settings. However, grounding is an interactive process in which people gradually develop shared understanding through ongoing communication. We introduce a four-metric suite (grounding efficiency, content alignment, lexical adaptation, and human-likeness) to systematically evaluate VLM performance in interactive grounding contexts. We deploy the suite on 150 self-play sessions of interactive referential games between three proprietary VLMs and compare them with human dyads. All three models diverge from human patterns on at least three metrics, while GPT4o-mini is the closest overall. We find that (i) task success scores do not indicate successful grounding and (ii) high image-utterance alignment does not necessarily predict task success. Our metric suite and findings offer a framework for future research on VLM grounding.

Evaluating sign language generation is often done through back-translation, where generated signs are first recognized back to text and then compared to a reference using text-based metrics. However, this two-step evaluation pipeline introduces ambiguity: it not only fails to capture the multimodal nature of sign language—such as facial expressions, spatial grammar, and prosody—but also makes it hard to pinpoint whether evaluation errors come from sign generation model or the translation system used to assess it. In this work, we propose SiLVERScore, a novel semantically-aware embedding-based evaluation metric that assesses sign language generation in a joint embedding space. Our contributions include: (1) identifying limitations of existing metrics, (2) introducing SiLVERScore for semantically-aware evaluation, (3) demonstrating its robustness to semantic and prosodic variations, and (4) exploring generalization challenges across datasets. On PHOENIX-14T and CSL-Daily datasets, SiLVERScore achieves near-perfect discrimination between correct and random pairs (ROC AUC = 0.99, overlap < 7%), substantially outperforming traditional metrics.

Large language models (LLMs) are capable of generating well-formed responses, but using LLMs to generate responses on the fly is not yet feasible for many task-oriented systems. Modular architectures are often still required for safety and privacy guarantees on the output. We hypothesize that an offline generation approach using discourse theories, formal grammar rules, and LLMs can allow us to generate human-like, coherent text in a more efficient, robust, and inclusive manner within a task-oriented setting. To this end, we present the first discourse-aware multimodal task-oriented dialogue system that combines discourse theories with offline LLM generation. We deploy our bot as an app to the general public and keep track of the user ratings for six months. Our user ratings show an improvement from 2.8 to 3.5 out of 5 with the introduction of discourse coherence theories. We also show that our model reduces misunderstandings in the dialect of African-American Vernacular English from 93% to 57%. While terms of use prevent us from releasing our entire codebase, we release our code in a format that can be integrated into most existing dialogue systems.