In this work we present the Social Influence Technique Taxonomy (SITT), a comprehensive framework of 58 empirically grounded techniques organized into nine categories, designed to detect subtle forms of social influence in textual content. We also investigate the LLMs ability to identify various forms of social influence. Building on interdisciplinary foundations, we construct the SITT dataset -- a 746-dialogue corpus annotated by 11 experts in Polish and translated into English -- to evaluate the ability of LLMs to identify these techniques. Using a hierarchical multi-label classification setup, we benchmark five LLMs, including GPT-4o, Claude 3.5, Llama-3.1, Mixtral, and PLLuM. Our results show that while some models, notably Claude 3.5, achieved moderate success (F1 score = 0.45 for categories), overall performance of models remains limited, particularly for context-sensitive techniques. The findings demonstrate key limitations in current LLMs' sensitivity to nuanced linguistic cues and underscore the importance of domain-specific fine-tuning. This work contributes a novel resource and evaluation example for understanding how LLMs detect, classify, and potentially replicate strategies of social influence in natural dialogues.

Lately I've been playing a new job sim game, Crime Scene Cleaner, while also watching BBC's comedy series The Cleaner, both of which focus on the aftermath of gruesome murders – sometimes you just need some cosy viewing to take the edge off the day. In the TV show, Greg Davies plays Wicky, the acerbic employee of a government-endorsed clean-up company, while Crime Scene Cleaner's lead character Kovalsky is a lowly janitor, mopping up blood and disposing of trash to cover up for a mob boss named Big Jim. The crime scenes in both are laughably over the top. I've never actually seen a real-life murder scene, so perhaps copious blood sprayed over walls and ceilings and the masses of broken furniture is completely normal. Stepping into Kovalsky's plastic overshoes, the aim is to leave each location exactly as it was prior to the … um … incident.

Kurt Knutsson talks about a viral video that shows a humanoid robot going wild. A chilling video circulating on social media has reignited old anxieties about robots turning against their creators. The footage shows a Unitree H1 humanoid robot, a machine about the size of an adult human, suddenly flailing its arms and legs with alarming force during a test, coming dangerously close to two technicians. The scene has sparked heated debate about the safety of advanced robotics and artificial intelligence. But is this truly the beginning of something out of our worst fears, or is there just a straightforward technical explanation for what happened?

While multimodal data sources are increasingly available from real-world forecasting, most existing research remains on unimodal time series. In this work, we present MoTime, a suite of multimodal time series forecasting datasets that pair temporal signals with external modalities such as text, metadata, and images. Covering diverse domains, MoTime supports structured evaluation of modality utility under two scenarios: 1) the common forecasting task, where varying-length history is available, and 2) cold-start forecasting, where no historical data is available. Experiments show that external modalities can improve forecasting performance in both scenarios, with particularly strong benefits for short series in some datasets, though the impact varies depending on data characteristics. By making datasets and findings publicly available, we aim to support more comprehensive and realistic benchmarks in future multimodal time series forecasting research.

This paper introduces an unsupervised framework for detecting audio patterns in musical samples (loops) through anomaly detection techniques, addressing challenges in music information retrieval (MIR). Existing methods are often constrained by reliance on handcrafted features, domain-specific limitations, or dependence on iterative user interaction. We address these limitations through an architecture combining deep feature extraction with unsupervised anomaly detection. Our approach leverages a pre-trained Hierarchical Token-semantic Audio Transformer (HTS-AT), paired with a Feature Fusion Mechanism (FFM), to generate representations from variable-length audio loops. These embeddings are processed using one-class Deep Support Vector Data Description (Deep SVDD), which learns normative audio patterns by mapping them to a compact latent hypersphere. Evaluations on curated bass and guitar datasets compare standard and residual autoencoder variants against baselines like Isolation Forest (IF) and and principle component analysis (PCA) methods. Results show our Deep SVDD models, especially the residual autoencoder variant, deliver improved anomaly separation, particularly for larger variations. This research contributes a flexible, fully unsupervised solution for processing diverse audio samples, overcoming previous structural and input limitations while enabling effective pattern identification through distance-based latent space scoring.

The increasing use of LLMs as substitutes for humans in ``aligning'' LLMs has raised questions about their ability to replicate human judgments and preferences, especially in ambivalent scenarios where humans disagree. This study examines the biases and limitations of LLMs in three roles: answer generator, judge, and debater. These roles loosely correspond to previously described alignment frameworks: preference alignment (judge) and scalable oversight (debater), with the answer generator reflecting the typical setting with user interactions. We develop a ``no-consensus'' benchmark by curating examples that encompass a variety of a priori ambivalent scenarios, each presenting two possible stances. Our results show that while LLMs can provide nuanced assessments when generating open-ended answers, they tend to take a stance on no-consensus topics when employed as judges or debaters. These findings underscore the necessity for more sophisticated methods for aligning LLMs without human oversight, highlighting that LLMs cannot fully capture human disagreement even on topics where humans themselves are divided.

The rapid evolution of social media has generated an overwhelming volume of user-generated content, conveying implicit opinions and contributing to the spread of misinformation. The method aims to enhance the detection of stance where misinformation can polarize user opinions. Stance detection has emerged as a crucial approach to effectively analyze underlying biases in shared information and combating misinformation. This paper proposes a novel method for \textbf{S}tance \textbf{P}rediction through a \textbf{L}abel-fused dual cross-\textbf{A}ttentive \textbf{E}motion-aware neural \textbf{Net}work (SPLAENet) in misinformative social media user-generated content. The proposed method employs a dual cross-attention mechanism and a hierarchical attention network to capture inter and intra-relationships by focusing on the relevant parts of source text in the context of reply text and vice versa. We incorporate emotions to effectively distinguish between different stance categories by leveraging the emotional alignment or divergence between the texts. We also employ label fusion that uses distance-metric learning to align extracted features with stance labels, improving the method's ability to accurately distinguish between stances. Extensive experiments demonstrate the significant improvements achieved by SPLAENet over existing state-of-the-art methods. SPLAENet demonstrates an average gain of 8.92\% in accuracy and 17.36\% in F1-score on the RumourEval dataset. On the SemEval dataset, it achieves average gains of 7.02\% in accuracy and 10.92\% in F1-score. On the P-stance dataset, it demonstrates average gains of 10.03\% in accuracy and 11.18\% in F1-score. These results validate the effectiveness of the proposed method for stance detection in the context of misinformative social media content.

The widespread success of large language models (LLMs) on NLP benchmarks has been accompanied by concerns that LLMs function primarily as stochastic parrots that reproduce texts similar to what they saw during pre-training, often erroneously. But what is the nature of their errors, and do these errors exhibit any regularities? In this work, we examine irrelevant context hallucinations, in which models integrate misleading contextual cues into their predictions. Through behavioral analysis, we show that these errors result from a structured yet flawed mechanism that we term class-based (mis)generalization, in which models combine abstract class cues with features extracted from the query or context to derive answers. Furthermore, mechanistic interpretability experiments on Llama-3, Mistral, and Pythia across 39 factual recall relation types reveal that this behavior is reflected in the model's internal computations: (i) abstract class representations are constructed in lower layers before being refined into specific answers in higher layers, (ii) feature selection is governed by two competing circuits -- one prioritizing direct query-based reasoning, the other incorporating contextual cues -- whose relative influences determine the final output. Our findings provide a more nuanced perspective on the stochastic parrot argument: through form-based training, LLMs can exhibit generalization leveraging abstractions, albeit in unreliable ways based on contextual cues -- what we term stochastic chameleons.

Large language models (LLMs) exhibit excellent performance in natural language processing (NLP), but remain highly sensitive to the quality of input queries, especially when these queries contain misleading or inaccurate information. Existing methods focus on correcting the output, but they often overlook the potential of improving the ability of LLMs to detect and correct misleading content in the input itself. In this paper, we propose a novel three-stage fine-tuning method that enhances the ability of LLMs to detect and correct misleading information in the input, further improving response accuracy and reducing hallucinations. Specifically, the three stages include (1) training LLMs to identify misleading information, (2) training LLMs to correct the misleading information using built-in or external knowledge, and (3) training LLMs to generate accurate answers based on the corrected queries. To evaluate our method, we conducted experiments on three datasets for the hallucination detection task and the question answering~(QA) task, as well as two datasets containing misleading information that we constructed. The experimental results demonstrate that our method significantly improves the accuracy and factuality of LLM responses, while also enhancing the ability to detect hallucinations and reducing the generation of hallucinations in the output, particularly when the query contains misleading information.

Endowing dialogue agents with persona information has proven to significantly improve the consistency and diversity of their generations. While much focus has been placed on aligning dialogues with provided personas, the adaptation to the interlocutor's profile remains largely underexplored. In this work, we investigate three key aspects: (1) a model's ability to align responses with both the provided persona and the interlocutor's; (2) its robustness when dealing with familiar versus unfamiliar interlocutors and topics, and (3) the impact of additional fine-tuning on specific persona-based dialogues. We evaluate dialogues generated with diverse speaker pairings and topics, framing the evaluation as an author identification task and employing both LLM-as-a-judge and human evaluations. By systematically masking or disclosing information about the interlocutor, we assess its impact on dialogue generation. Results show that access to the interlocutor's persona improves the recognition of the target speaker, while masking it does the opposite. Although models generalise well across topics, they struggle with unfamiliar interlocutors. Finally, we found that in zero-shot settings, LLMs often copy biographical details, facilitating identification but trivialising the task.