Papers by Kaitlyn Zhou
ReasonIF: Large Reasoning Models Fail to Follow Instructions During Reasoning (2026.findings-acl)
Copied to clipboard
| Challenge: | Prior studies assess instruction adherence in the model’s main responses, but it is also critical for large reasoning models to follow user instructions throughout their reasoning process. |
| Approach: | They propose a systematic benchmark for assessing reasoning instruction following to assess the model's adherence to instructions. |
| Outcome: | The proposed benchmark reduces the risk of undesirable shortcuts, hallucinations, or reward hacking within reasoning traces. |
Rethinking Word Similarity: Semantic Similarity through Classification Confusion (2025.naacl-long)
Copied to clipboard
| Challenge: | Word similarity measures cannot capture context-dependent, asymmetrical, polysemous nature of semantic similarity. |
| Approach: | They propose a new measure of similarity that reframes semantic similarity in terms of feature-based classification confusion. |
| Outcome: | The proposed model is comparable to cosine similarity in matching human similarity judgments across several datasets and can measure similarity using predetermined features of interest. |
Deconstructing NLG Evaluation: Evaluation Practices, Assumptions, and Their Implications (2022.naacl-main)
Copied to clipboard
| Challenge: | Evaluating natural language generation systems is difficult, as there are many ways to express similar things in text. |
| Approach: | They combine interviews with NLG practitioners to examine ethical considerations and their implications for NLG evaluation. |
| Outcome: | The findings of the study surface goals, community practices, assumptions, and constraints that shape NLG evaluations, and examine their implications and how they embody ethical considerations. |
Navigating the Grey Area: How Expressions of Uncertainty and Overconfidence Affect Language Models (2023.emnlp-main)
Copied to clipboard
| Challenge: | a recent study examines how epistemic markers of certainty, uncertainty, or evidentiality affect LMs' attitudes toward knowledge . accuracies of expressions of high certainty and low certainty are associated with questions . prior work focused on learning the mapping between internal probabilities of a model and an ordinal output . |
| Approach: | They develop a typology of epistemic markers and inject 50 markers into questions . they find that LMs are highly sensitive to epistemical markers in prompts . |
| Outcome: | The proposed model epistemology study shows that LMs are sensitive to epistemic markers in prompts . expressions of high certainty result in 7% decrease in accuracy, while factive verbs hurt performance . |
Problems with Cosine as a Measure of Embedding Similarity for High Frequency Words (2022.acl-short)
Copied to clipboard
| Challenge: | We find that word similarities estimated by cosine over contextual embeddings are understated and trace this effect to training data frequency. |
| Approach: | They propose to use cosine similarity to estimate word similarities in contextual embeddings to trace this effect to training data frequency. |
| Outcome: | The proposed model underestimates similarity between frequent and low frequency words even after controlling for polysemy and other factors. |
Relying on the Unreliable: The Impact of Language Models’ Reluctance to Express Uncertainty (2024.acl-long)
Copied to clipboard
| Challenge: | a pivotal aspect of fostering reliable human-AI interactions lies in the apt communication of model confidences. |
| Approach: | They examine how LMs incorporate confidence in responses via natural language . they also examine how downstream users behave in response to LM-articulated uncertainties . |
| Outcome: | The proposed model overconfidences are high in LMs, and humans are biased against uncertainty-rich texts. |
Richer Countries and Richer Representations (2022.findings-acl)
Copied to clipboard
| Challenge: | Using BERT, countries with low frequency in training data are less likely to be invocabulary, and are less frequently predicted in the masked language modeling task. |
| Approach: | They propose three criteria to characterize the quality of representations for particular entities or groups: consistency, distinctiveness, and recognizability. |
| Outcome: | The results suggest that frequency is highly correlated with a country’s GDP, perpetuating historic power and wealth inequalities. |
ELI-Why: Evaluating the Pedagogical Utility of Language Model Explanations (2025.findings-acl)
Copied to clipboard
Brihi Joshi, Keyu He, Sahana Ramnath, Sadra Sabouri, Kaitlyn Zhou, Souti Chattopadhyay, Swabha Swayamdipta, Xiang Ren
| Challenge: | Language models are widely used in education, yet their ability to tailor responses to learners with varied informational needs and knowledge backgrounds remains under-explored. |
| Approach: | They conduct two extensive human studies to assess the utility of language model-generated explanatory answers (explanations) on a benchmark of 13.4K "Why" questions. |
| Outcome: | The proposed model explanations match learners' educational backgrounds only 50% of the time, compared to 79% for lay explanations. |
REL-A.I.: An Interaction-Centered Approach To Measuring Human-LM Reliance (2025.naacl-long)
Copied to clipboard
| Challenge: | Existing evaluations of large language models' ability to communicate uncertainty and knowledge limitations focus on the behaviors of their human interlocutors. |
| Approach: | They propose an interaction-centered evaluation approach that quantifies whether and how humans rely on LLMs' responses. |
| Outcome: | The proposed approach quantifies whether and how humans rely on LLMs' responses. |
Attention to Non-Adopters (2026.findings-acl)
Copied to clipboard
Kaitlyn Zhou, Kristina Gligorić, Myra Cheng, Michelle S. Lam, Vyoma Raman, Boluwatife Aminu, Caeley Woo, Michael Brockman, Hannah Cha, Dan Jurafsky
| Challenge: | incorporating non-adopter perspectives is essential for developing useful and capable LLMs, argues a new study. |
| Approach: | They argue that incorporating non-adopter perspectives is essential for developing broadly useful and capable LLMs. |
| Outcome: | The proposed method will risk missing tasks prioritized by non-adopters, the authors argue . they show that non-dots diverge from those of current users, and non-no-acopter needs point towards novel reasoning tasks. |