With the alarming rise in active shooter incidents (ASIs) in the United States, enhancing public safety through building design has become a pressing need. This study proposes a reinforcement learning-based simulation approach addressing gaps in existing research that has neglected the dynamic behaviours of shooters. We developed an autonomous agent to simulate an active shooter within a realistic office environment, aiming to offer insights into the interactions between building design parameters and ASI outcomes. A case study is conducted to quantitatively investigate the impact of building exit numbers (total count of accessible exits) and configuration (arrangement of which exits are available or not) on evacuation and harm rates. Findings demonstrate that greater exit availability significantly improves evacuation outcomes and reduces harm. Exits nearer to the shooter's initial position hold greater importance for accessibility than those farther away. By encompassing dynamic shooter behaviours, this study offers preliminary insights into effective building safety design against evolving threats.

We present a comprehensive evaluation of proprietary and open-weights large language models using the first astronomy-specific benchmarking dataset. This dataset comprises 4,425 multiple-choice questions curated from the Annual Review of Astronomy and Astrophysics, covering a broad range of astrophysical topics. Our analysis examines model performance across various astronomical subfields and assesses response calibration, crucial for potential deployment in research environments. Claude-3.5-Sonnet outperforms competitors by up to 4.6 percentage points, achieving 85.0% accuracy. For proprietary models, we observed a universal reduction in cost every 3-to-12 months to achieve similar score in this particular astronomy benchmark. Open-source models have rapidly improved, with LLaMA-3-70b (80.6%) and Qwen-2-72b (77.7%) now competing with some of the best proprietary models. We identify performance variations across topics, with non-English-focused models generally struggling more in exoplanet-related fields, stellar astrophysics, and instrumentation related questions. These challenges likely stem from less abundant training data, limited historical context, and rapid recent developments in these areas. This pattern is observed across both open-weights and proprietary models, with regional dependencies evident, highlighting the impact of training data diversity on model performance in specialized scientific domains. Top-performing models demonstrate well-calibrated confidence, with correlations above 0.9 between confidence and correctness, though they tend to be slightly underconfident. The development for fast, low-cost inference of open-weights models presents new opportunities for affordable deployment in astronomy. The rapid progress observed suggests that LLM-driven research in astronomy may become feasible in the near future.

Codebooks -- documents that operationalize constructs and outline annotation procedures -- are used almost universally by social scientists when coding unstructured political texts. Recently, to reduce manual annotation costs, political scientists have looked to generative large language models (LLMs) to label and analyze text data. However, previous work using LLMs for classification has implicitly relied on the universal label assumption -- correct classification of documents is possible using only a class label or minimal definition and the information that the LLM inductively learns during its pre-training. In contrast, we argue that political scientists who care about valid measurement should instead make a codebook-construct label assumption -- an LLM should follow the definition and exclusion criteria of a construct/label provided in a codebook. In this work, we collect and curate three political science datasets and their original codebooks and conduct a set of experiments to understand whether LLMs comply with codebook instructions, whether rewriting codebooks improves performance, and whether instruction-tuning LLMs on codebook-document-label tuples improves performance over zero-shot classification. Using Mistral 7B Instruct as our LLM, we find re-structuring the original codebooks gives modest gains in zero-shot performance but the model still struggles to comply with the constraints of the codebooks. Optimistically, instruction-tuning Mistral on one of our datasets gives significant gains over zero-shot inference (0.76 versus 0.53 micro F1). We hope our conceptualization of the codebook-specific task, assumptions, and instruction-tuning pipeline as well our semi-structured LLM codebook format will help political scientists readily adapt to the LLM era.

With rapid advances in machine learning, many people in the field have been discussing the rise of digital minds and the possibility of artificial sentience. Future developments in AI capabilities and safety will depend on public opinion and human-AI interaction. To begin to fill this research gap, we present the first nationally representative survey data on the topic of sentient AI: initial results from the Artificial Intelligence, Morality, and Sentience (AIMS) survey, a preregistered and longitudinal study of U.S. public opinion that began in 2021. Across one wave of data collection in 2021 and two in 2023 (total N = 3,500), we found mind perception and moral concern for AI well-being in 2021 were higher than predicted and significantly increased in 2023: for example, 71% agree sentient AI deserve to be treated with respect, and 38% support legal rights. People have become more threatened by AI, and there is widespread opposition to new technologies: 63% support a ban on smarter-than-human AI, and 69% support a ban on sentient AI. Expected timelines are surprisingly short and shortening with a median forecast of sentient AI in only five years and artificial general intelligence in only two years. We argue that, whether or not AIs become sentient, the discussion itself may overhaul human-computer interaction and shape the future trajectory of AI technologies, including existential risks and opportunities.

Recently, code-oriented large language models (Code LLMs) have been widely and successfully used to simplify and facilitate code programming. With these tools, developers can easily generate desired complete functional codes based on incomplete code and natural language prompts. However, a few pioneering works revealed that these Code LLMs are also vulnerable, e.g., against backdoor and adversarial attacks. The former could induce LLMs to respond to triggers to insert malicious code snippets by poisoning the training data or model parameters, while the latter can craft malicious adversarial input codes to reduce the quality of generated codes. However, both attack methods have underlying limitations: backdoor attacks rely on controlling the model training process, while adversarial attacks struggle with fulfilling specific malicious purposes. To inherit the advantages of both backdoor and adversarial attacks, this paper proposes a new attack paradigm, i.e., target-specific and adversarial prompt injection (TAPI), against Code LLMs. TAPI generates unreadable comments containing information about malicious instructions and hides them as triggers in the external source code. When users exploit Code LLMs to complete codes containing the trigger, the models will generate attacker-specified malicious code snippets at specific locations. We evaluate our TAPI attack on four representative LLMs under three representative malicious objectives and seven cases. The results show that our method is highly threatening (achieving an attack success rate enhancement of up to 89.3%) and stealthy (saving an average of 53.1% of tokens in the trigger design). In particular, we successfully attack some famous deployed code completion integrated applications, including CodeGeex and Github Copilot. This further confirms the realistic threat of our attack.

Air pollution remains one of the most formidable environmental threats to human health globally, particularly in urban areas, contributing to nearly 7 million premature deaths annually. Megacities, defined as cities with populations exceeding 10 million, are frequent hotspots of severe pollution, experiencing numerous weeks of dangerously poor air quality due to the concentration of harmful pollutants. In addition, the complex interplay of factors makes accurate air quality predictions incredibly challenging, and prediction models often struggle to capture these intricate dynamics. To address these challenges, this paper proposes an attention-enhanced deep multitask spatiotemporal machine learning model based on long-short-term memory networks for long-term air quality monitoring and prediction. The model demonstrates robust performance in predicting the levels of major pollutants such as sulfur dioxide and carbon monoxide, effectively capturing complex trends and fluctuations. The proposed model provides actionable information for policymakers, enabling informed decision making to improve urban air quality.

Conversational large language models are fine-tuned for both instruction-following and safety, resulting in models that obey benign requests but refuse harmful ones. While this refusal behavior is widespread across chat models, its underlying mechanisms remain poorly understood. In this work, we show that refusal is mediated by a one-dimensional subspace, across 13 popular open-source chat models up to 72B parameters in size. Specifically, for each model, we find a single direction such that erasing this direction from the model's residual stream activations prevents it from refusing harmful instructions, while adding this direction elicits refusal on even harmless instructions. Leveraging this insight, we propose a novel white-box jailbreak method that surgically disables refusal with minimal effect on other capabilities. Finally, we mechanistically analyze how adversarial suffixes suppress propagation of the refusal-mediating direction. Our findings underscore the brittleness of current safety fine-tuning methods. More broadly, our work showcases how an understanding of model internals can be leveraged to develop practical methods for controlling model behavior.

Personal names simultaneously differentiate individuals and categorize them in ways that are important in a given society. While the natural language processing community has thus associated personal names with sociodemographic characteristics in a variety of tasks, researchers have engaged to varying degrees with the established methodological problems in doing so. To guide future work that uses names and sociodemographic characteristics, we provide an overview of relevant research: first, we present an interdisciplinary background on names and naming. We then survey the issues inherent to associating names with sociodemographic attributes, covering problems of validity (e.g., systematic error, construct validity), as well as ethical concerns (e.g., harms, differential impact, cultural insensitivity). Finally, we provide guiding questions along with normative recommendations to avoid validity and ethical pitfalls when dealing with names and sociodemographic characteristics in natural language processing.

Entity Linking (EL) models are well-trained at mapping mentions to their corresponding entities according to a given context. However, EL models struggle to disambiguate long-tail entities due to their limited training data. Meanwhile, large language models (LLMs) are more robust at interpreting uncommon mentions. Yet, due to a lack of specialized training, LLMs suffer at generating correct entity IDs. Furthermore, training an LLM to perform EL is cost-intensive. Building upon these insights, we introduce LLM-Augmented Entity Linking LLMAEL, a plug-and-play approach to enhance entity linking through LLM data augmentation. We leverage LLMs as knowledgeable context augmenters, generating mention-centered descriptions as additional input, while preserving traditional EL models for task specific processing. Experiments on 6 standard datasets show that the vanilla LLMAEL outperforms baseline EL models in most cases, while the fine-tuned LLMAEL set the new state-of-the-art results across all 6 benchmarks.

Compositional generalization is an important ability of language models and has many different manifestations. For data-to-text generation, previous research on this ability is limited to a single manifestation called Systematicity and lacks consideration of large language models (LLMs), which cannot fully cover practical application scenarios. In this work, we propose SPOR, a comprehensive and practical evaluation method for compositional generalization in data-to-text generation. SPOR includes four aspects of manifestations (Systematicity, Productivity, Order invariance, and Rule learnability) and allows high-quality evaluation without additional manual annotations based on existing datasets. We demonstrate SPOR on two different datasets and evaluate some existing language models including LLMs. We find that the models are deficient in various aspects of the evaluation and need further improvement. Our work shows the necessity for comprehensive research on different manifestations of compositional generalization in data-to-text generation and provides a framework for evaluation.