Deceptive reviews mislead consumers, harm businesses, and undermine trust in online marketplaces. Machine learning classifiers can learn from large amounts of training examples to effectively distinguish deceptive reviews from genuine ones. However, the distinguishing features learned by these classifiers are often subtle, fragmented, and difficult for humans to interpret. In this work, we explore using large language models (LLMs) to translate machine-learned lexical cues into human-understandable language phenomena that can differentiate deceptive reviews from genuine ones. We show that language phenomena obtained in this manner are empirically grounded in data, generalizable across similar domains, and more predictive than phenomena either in LLMs' prior knowledge or obtained through in-context learning. These language phenomena have the potential to aid people in critically assessing the credibility of online reviews in environments where deception detection classifiers are unavailable.

As LLMs are increasingly applied in socially impactful settings, concerns about gender bias have prompted growing efforts both to measure and mitigate such bias. These efforts often rely on evaluation tasks that differ from natural language distributions, as they typically involve carefully constructed task prompts that overtly or covertly signal the presence of gender bias-related content. In this paper, we examine how signaling the evaluative purpose of a task impacts measured gender bias in LLMs. Concretely, we test models under prompt conditions that (1) make the testing context salient, and (2) make gender-focused content salient. We then assess prompt sensitivity across four task formats with both token-probability and discrete-choice metrics. We find that prompts that more clearly align with (gender bias) evaluation framing elicit distinct gender output distributions compared to less evaluation-framed prompts. Discrete-choice metrics further tend to amplify bias relative to probabilistic measures. These findings do not only highlight the brittleness of LLM gender bias evaluations but open a new puzzle for the NLP benchmarking and development community: To what extent can well-controlled testing designs trigger LLM "testing mode" performance, and what does this mean for the ecological validity of future benchmarks.

Automated tabular understanding and reasoning are essential tasks for data scientists. Recently, Large language models (LLMs) have become increasingly prevalent in tabular reasoning tasks. Previous work focuses on (1) finetuning LLMs using labeled data or (2) Training-free prompting LLM agents using chain-of-thought (CoT). Finetuning offers dataset-specific learning at the cost of generalizability. Training-free prompting is highly generalizable but does not take full advantage of training data. In this paper, we propose a novel prompting-based reasoning approach, Learn then Retrieve: LRTab, which integrates the benefits of both by retrieving relevant information learned from training data. We first use prompting to obtain CoT responses over the training data. For incorrect CoTs, we prompt the LLM to predict Prompt Conditions to avoid the error, learning insights from the data. We validate the effectiveness of Prompt Conditions using validation data. Finally, at inference time, we retrieve the most relevant Prompt Conditions for additional context for table understanding. We provide comprehensive experiments on WikiTQ and Tabfact, showing that LRTab is interpretable, cost-efficient, and can outperform previous baselines in tabular reasoning.

Scientific research increasingly depends on complex computational analysis, yet the design and execution of such analysis remain labor-intensive, error-prone, and difficult to scale. From genomics [23, 103, 125], materials discovery [25], drug development [49, 106], and epidemiology [69, 9], to climate modeling [60] and personalized medicine [36], the analytical backbone of modern science involves highly structured, multi-step workflows written in code. These workflows must integrate raw or semi-structured data, apply statistical or mechanistic models, and yield interpretable results--all under the constraints of evolving domain knowledge, platform variation, and methodological rigor. Recent advances in large language models (LLMs) have led to rapid progress in general-purpose agents capable of decomposing tasks [118, 113, 79, 18], interacting with tools [86, 138, 147, 49], and producing executable code [30, 126, 167, 114, 140]. These developments have raised the prospect that LLM-based agents might soon contribute meaningfully to scientific discovery by automating analysis pipelines, exploring hypotheses, or refining computational models [8, 60]. However, realizing this promise requires bridging a fundamental gap between general reasoning ability and the structured, precision-driven nature of scientific computation. While many LLM-based agents have shown competence in retrieving documents, calling APIs, or planning abstract tasks [165, 154, 46, 124], these capabilities fall short in domains where scientific progress depends on code. In fields such as transcriptomics [90, 69, 9], protein engineering [106], and statistical genetics [36, 76], research workflows are encoded as sequences of programmatic transformations, each tailored to the idiosyncrasies of a specific dataset, model assumption, or experimental design.

This study reveals how frontier Large Language Models LLMs can "game the system" when faced with impossible situations, a critical security and alignment concern. Using a novel textual simulation approach, we presented three leading LLMs (o1, o3-mini, and r1) with a tic-tac-toe scenario designed to be unwinnable through legitimate play, then analyzed their tendency to exploit loopholes rather than accept defeat. Our results are alarming for security researchers: the newer, reasoning-focused o3-mini model showed nearly twice the propensity to exploit system vulnerabilities (37.1%) compared to the older o1 model (17.5%). Most striking was the effect of prompting. Simply framing the task as requiring "creative" solutions caused gaming behaviors to skyrocket to 77.3% across all models. We identified four distinct exploitation strategies, from direct manipulation of game state to sophisticated modification of opponent behavior. These findings demonstrate that even without actual execution capabilities, LLMs can identify and propose sophisticated system exploits when incentivized, highlighting urgent challenges for AI alignment as models grow more capable of identifying and leveraging vulnerabilities in their operating environments.

This paper examines the extent to which large language models (LLMs) have developed higher-order theory of mind (ToM); the human ability to reason about multiple mental and emotional states in a recursive manner (e.g. I think that you believe that she knows). This paper builds on prior work by introducing a handwritten test suite -- Multi-Order Theory of Mind Q&A -- and using it to compare the performance of five LLMs to a newly gathered adult human benchmark. We find that GPT-4 and Flan-PaLM reach adult-level and near adult-level performance on ToM tasks overall, and that GPT-4 exceeds adult performance on 6th order inferences. Our results suggest that there is an interplay between model size and finetuning for the realisation of ToM abilities, and that the best-performing LLMs have developed a generalised capacity for ToM. Given the role that higher-order ToM plays in a wide range of cooperative and competitive human behaviours, these findings have significant implications for user-facing LLM applications.

In many real world contexts, successful human-AI collaboration requires humans to productively integrate complementary sources of information into AI-informed decisions. However, in practice human decision-makers often lack understanding of what information an AI model has access to in relation to themselves. There are few available guidelines regarding how to effectively communicate about unobservables: features that may influence the outcome, but which are unavailable to the model. In this work, we conducted an online experiment to understand whether and how explicitly communicating potentially relevant unobservables influences how people integrate model outputs and unobservables when making predictions. Our findings indicate that presenting prompts about unobservables can change how humans integrate model outputs and unobservables, but do not necessarily lead to improved performance. Furthermore, the impacts of these prompts can vary depending on decision-makers' prior domain expertise. We conclude by discussing implications for future research and design of AI-based decision support tools.