Mark Dredze

Large language models (LLMs) rely on both contextual knowledge and parametric memory, yet these sources can conflict. Prior analysis largely focused on contextual question answering, suggesting that models tend to favor parametric knowledge under conflict, but this setting assumes that tasks should always rely on the provided passage. It therefore remains unclear how LLMs behave when tasks demand different kinds and degrees of knowledge utilization. We address this gap with a model-agnostic diagnostic framework that holds underlying knowledge constant while injecting controlled conflicts across tasks with varying knowledge requirements. Evaluating representative open-source LLMs, we find that: (1) performance degradation under conflict correlates with a task's knowledge reliance rather than conflict plausibility alone; (2) strategies such as explanatory rationales or reiteration increase context reliance, helping context-only tasks but harming those that require parametric knowledge; and (3) these behaviors bias model-based evaluation, raising concerns about the reliability of LLMs as judges. Together, our findings show that context–memory conflict is fundamentally task-dependent and motivate task-aware approaches to balancing context and memory in LLM deployment and evaluation.

Artificial Intelligence (AI) holds immense promise for public health, yet its potential is undermined by alignment failures where systems act contrary to human values, often exacerbating health disparities. This paper challenges the narrow view that algorithmic bias is solely a data problem, arguing instead that misalignment arises at every stage of the AI development lifecycle. We introduce a comprehensive seven-stage framework, spanning problem definition, team assembly, study design, data acquisition, model training, validation, and post-deployment implementation, viewed through an epidemiological lens. This approach systematically integrates core principles such as population representativeness, rigorous study design, bias characterization, and causal reasoning to identify and mitigate alignment risks. For each stage, we define specific alignment failures, from flawed problem formulation to post-market performance degradation, and propose actionable, evidence-based solutions. By embedding epidemiological rigor throughout the entire AI lifecycle, this framework provides a structured, proactive pathway for researchers, developers, and policymakers to create trustworthy, safe, and fair AI systems. This systemic approach is critical to harnessing AI's transformative benefits for population health while preventing the perpetuation of inequity and harm.

The majority of work on summarization evaluation focuses on general summary quality (e.g., ROUGE, BERTScore) or specific desired properties (e.g., readability, factuality). However, these metrics fail to measure the utility of a summary to an individual user. For example, a biomedical researcher learning about the latest vaccine research will have different informational needs from a family doctor. Query-focused summarization captures part of this need, but in practice, users rarely state everything relevant in a query: a single short query is likely inadequate to distinguish the needs of a researcher from those of a physician. By contrast, a reader's background or persona (their role and expertise) is comparatively stable across queries and recovers much of this missing context, which makes it a practical signal for assessing whether a summary satisfies that reader's needs. In this work, we assess how sensitive popular summarization metrics are to both informational and persona differences, and find that many popular metrics, including strong LLM-as-judge metrics, fail basic perturbation tests of informational content. We additionally conduct an expert human evaluation, measuring summary preferences based on information satisfaction given a specific person's background and use case. We find that both traditional and LLM-based metrics are insufficient measures of information satisfaction and agree poorly with human judgment.

To responsibly develop Generative AI (GenAI) products, it is critical to define the scope of acceptable inputs and outputs. What constitutes a "safe" response is an actively debated question. Academic work puts an outsized focus on evaluating models by themselves for general purpose aspects such as toxicity, bias, and fairness, especially in conversational applications being used by a broad audience. In contrast, less focus is put on considering sociotechnical systems in specialized domains. Yet, those specialized systems can be subject to extensive and well-understood legal and regulatory scrutiny. These product-specific considerations need to be set in industry-specific laws, regulations, and corporate governance requirements. In this paper, we aim to highlight AI content safety considerations specific to the financial services domain and outline an associated AI content risk taxonomy. We compare this taxonomy to existing work in this space and discuss implications of risk category violations on various stakeholders. We evaluate how existing open-source technical guardrail solutions cover this taxonomy by assessing them on data collected via red-teaming activities. Our results demonstrate that these guardrails fail to detect most of the content risks we discuss.

Large language model development relies on the pre-train-then-align paradigm, in which the model is typically pre-trained on a large text corpus and undergoes a tuning stage to align the model with human preference or downstream tasks. We investigate the relationship between pre-training and supervised fine-tuning by considering multiple tasks as well as different pre-trained model checkpoints. Our results on 18 datasets and two models suggest that i) although the model benefits significantly through supervised fine-tuning, it may forget previously known domain knowledge and tasks that are not seen during fine-tuning; ii) the model exhibits high sensitivity to evaluation prompts after supervised fine-tuning, but this sensitivity can be alleviated through further pre-training; iii) continual pre-training improves the model in a latent way that manifests after fine-tuning; iv) The model can already solve some tasks after pre-training while fine-tuning most benefits datasets where the model does not show capability during pre-training.

Efforts to ensure the safety of large language models (LLMs) include safety fine-tuning, evaluation, and red teaming. However, despite the widespread use of the Retrieval-Augmented Generation (RAG) framework, AI safety work focuses on standard LLMs, which means we know little about how RAG use cases change a model's safety profile. We conduct a detailed comparative analysis of RAG and non-RAG frameworks with eleven LLMs. We find that RAG can make models less safe and change their safety profile. We explore the causes of this change and find that even combinations of safe models with safe documents can cause unsafe generations. In addition, we evaluate some existing red teaming methods for RAG settings and show that they are less effective than when used for non-RAG settings. Our work highlights the need for safety research and red-teaming methods specifically tailored for RAG LLMs.

Autonomous agents such as Claude Code and Codex now operate for hours or even days. Understanding their runtime behavior has become critical for downstream tasks such as diagnosing inefficiencies, fixing bugs, and ensuring better oversight. A primary way to gain this understanding is analyzing the reasoning trajectories and execution traces these agents generate. Yet such data remains in unstructured natural-language form, making it difficult for humans to interpret at scale. We introduce ACT*ONOMY (a combination of Action and Taxonomy), a taxonomy for describing and analyzing agent behavior at runtime. ACT*ONOMY has two components: (1) the taxonomy itself, developed through Grounded Theory and structured as a three-level hierarchy of 10 actions, 46 subactions, and 120 leaf categories; and (2) an open repository that hosts the living taxonomy, provides an automated analysis pipeline that applies it to agent trajectories analysis, and defines an extension protocol for customization and growth. Our experiments show that ACTONOMY can compare behavioral profiles across agents and characterize a single agent's behavior across diverse trajectories, surfacing patterns indicative of failure modes. By providing a shared vocabulary, ACT*ONOMY helps researchers, agent designers, and end users interpret agent behavior more consistently, enabling better oversight and control.