We propose a novel preference modeling approach that utilizes AI feedback and only requires a small amount of human data.

Computational humor is a frontier for creating advanced and engaging natural language processing (NLP) applications, such as sophisticated dialogue systems. While previous studies have benchmarked the humor capabilities of Large Language Models (LLMs), they have often relied on single-dimensional evaluations, such as judging whether something is simply ``funny.'' This paper argues that a multifaceted understanding of humor is necessary and addresses this gap by systematically evaluating LLMs through the lens of Oogiri, a form of Japanese improvisational comedy games. To achieve this, we expanded upon existing Oogiri datasets with data from new sources and then augmented the collection with Oogiri responses generated by LLMs. We then manually annotated this expanded collection with 5-point absolute ratings across six dimensions: Novelty, Clarity, Relevance, Intelligence, Empathy, and Overall Funniness. Using this dataset, we assessed the capabilities of state-of-the-art LLMs on two core tasks: their ability to generate creative Oogiri responses and their ability to evaluate the funniness of responses using a six-dimensional evaluation. Our results show that while LLMs can generate responses at a level between low- and mid-tier human performance, they exhibit a notable lack of Empathy. This deficit in Empathy helps explain their failure to replicate human humor assessment. Correlation analyses of human and model evaluation data further reveal a fundamental divergence in evaluation criteria: LLMs prioritize Novelty, whereas humans prioritize Empathy. We release our annotated corpus to the community to pave the way for the development of more emotionally intelligent and sophisticated conversational agents.

Large Language Models (LLMs) often expend significant computational resources generating boilerplate responses, such as refusals, simple acknowledgements and casual greetings, which adds unnecessary cost and latency. To address this inefficiency, we propose a simple yet highly effective method for detecting such responses after only a single generation step. We demonstrate that the log-probability distribution of the first generated token serves as a powerful signal for classifying the nature of the entire subsequent response. Our experiments, conducted across a diverse range of small, large, and reasoning-specialized models, show that the first-token log-probability vectors form distinctly separable clusters for different response types. Using a lightweight k-NN classifier, we achieve high accuracy in predicting whether a response will be a substantive answer or a form of boilerplate response, including user-specified refusals. The primary implication is a practical, computationally trivial technique, optimizing LLM inference by enabling early termination or redirection to a smaller model, thereby yielding significant savings in computational cost. This work presents a direct path toward more efficient and sustainable LLM deployment.

We propose a novel preference modeling approach that utilizes AI feedback and only requires a small amount of human data.

Interactive intelligent agents are being integrated across society. Despite achieving human-like capabilities, humans' responses to these agents remain poorly understood, with research fragmented across disciplines. We conducted a first systematic synthesis comparing a range of psychological and behavioural responses in matched human-agent vs. human-human dyadic interactions. A total of 162 eligible studies (146 contributed to the meta-analysis; 468 effect sizes) were included in the systematic review and meta-analysis, which integrated frequentist and Bayesian approaches. Our results indicate that individuals exhibited less prosocial behaviour and moral engagement when interacting with agents vs. humans. They attributed less agency and responsibility to agents, perceiving them as less competent, likeable, and socially present. In contrast, individuals' social alignment (i.e., alignment or adaptation of internal states and behaviours with partners), trust in partners, personal agency, task performance, and interaction experiences were generally comparable when interacting with agents vs. humans. We observed high effect-size heterogeneity for many subjective responses (i.e., social perceptions of partners, subjective trust, and interaction experiences), suggesting context-dependency of partner effects. By examining the characteristics of studies, participants, partners, interaction scenarios, and response measures, we also identified several moderators shaping partner effects. Overall, functional behaviours and interactive experiences with agents can resemble those with humans, whereas fundamental social attributions and moral/prosocial concerns lag in human-agent interactions. Agents are thus afforded instrumental value on par with humans but lack comparable intrinsic value, providing practical implications for agent design and regulation.

While large language model (LLM)-based chatbots have demonstrated strong capabilities in generating coherent and contextually relevant responses, they often struggle with understanding when to speak, particularly in delivering brief, timely reactions during ongoing conversations. This limitation arises largely from their reliance on text input, lacking the rich contextual cues in real-world human dialogue. In this work, we focus on real-time prediction of response types, with an emphasis on short, reactive utterances that depend on subtle, multimodal signals across vision, audio, and text. To support this, we introduce a new multimodal dataset constructed from real-world conversational videos, containing temporally aligned visual, auditory, and textual streams. This dataset enables fine-grained modeling of response timing in dyadic interactions. Building on this dataset, we propose MM-When2Speak, a multimodal LLM-based model that adaptively integrates visual, auditory, and textual context to predict when a response should occur, and what type of response is appropriate. Experiments show that MM-When2Speak significantly outperforms state-of-the-art unimodal and LLM-based baselines, achieving up to a 4x improvement in response timing accuracy over leading commercial LLMs. These results underscore the importance of multimodal inputs for producing timely, natural, and engaging conversational AI.

AI assistants can impart value judgments that shape people's decisions and worldviews, yet little is known empirically about what values these systems rely on in practice. To address this, we develop a bottom-up, privacy-preserving method to extract the values (normative considerations stated or demonstrated in model responses) that Claude 3 and 3.5 models exhibit in hundreds of thousands of real-world interactions. We empirically discover and taxonomize 3,307 AI values and study how they vary by context. We find that Claude expresses many practical and epistemic values, and typically supports prosocial human values while resisting values like "moral nihilism". While some values appear consistently across contexts (e.g. "transparency"), many are more specialized and context-dependent, reflecting the diversity of human interlocutors and their varied contexts. For example, "harm prevention" emerges when Claude resists users, "historical accuracy" when responding to queries about controversial events, "healthy boundaries" when asked for relationship advice, and "human agency" in technology ethics discussions. By providing the first large-scale empirical mapping of AI values in deployment, our work creates a foundation for more grounded evaluation and design of values in AI systems.

Reinforcement learning based fine-tuning of large language models (LLMs) on human preferences has been shown to enhance both their capabilities and safety behavior. However, in cases related to safety, without precise instructions to human annotators, the data collected may cause the model to become overly cautious, or to respond in an undesirable style, such as being judgmental. Additionally, as model capabilities and usage patterns evolve, there may be a costly need to add or relabel data to modify safety behavior. We propose a novel preference modeling approach that utilizes AI feedback and only requires a small amount of human data. Our method, Rule Based Rewards (RBR), uses a collection of rules for desired or undesired behaviors (e.g. refusals should not be judgmental) along with a LLM grader. In contrast to prior methods using AI feedback, our method uses fine-grained, composable, LLM-graded few-shot prompts as reward directly in RL training, resulting in greater control, accuracy and ease of updating. We show that RBRs are an effective training method, achieving an F1 score of 97.1, compared to a human-feedback baseline of 91.7, resulting in much higher safety-behavior accuracy through better balancing usefulness and safety.

This paper presents a comprehensive analysis of how AI-generated medical responses are perceived and evaluated by non-experts. A total of 300 participants gave evaluations for medical responses that were either written by a medical doctor on an online healthcare platform, or generated by a large language model and labeled by physicians as having high or low accuracy. Results showed that participants could not effectively distinguish between AI-generated and Doctors' responses and demonstrated a preference for AI-generated responses, rating High Accuracy AI-generated responses as significantly more valid, trustworthy, and complete/satisfactory. Low Accuracy AI-generated responses on average performed very similar to Doctors' responses, if not more. Participants not only found these low-accuracy AI-generated responses to be valid, trustworthy, and complete/satisfactory but also indicated a high tendency to follow the potentially harmful medical advice and incorrectly seek unnecessary medical attention as a result of the response provided. This problematic reaction was comparable if not more to the reaction they displayed towards doctors' responses. This increased trust placed on inaccurate or inappropriate AI-generated medical advice can lead to misdiagnosis and harmful consequences for individuals seeking help. Further, participants were more trusting of High Accuracy AI-generated responses when told they were given by a doctor and experts rated AI-generated responses significantly higher when the source of the response was unknown. Both experts and non-experts exhibited bias, finding AI-generated responses to be more thorough and accurate than Doctors' responses but still valuing the involvement of a Doctor in the delivery of their medical advice. Ensuring AI systems are implemented with medical professionals should be the future of using AI for the delivery of medical advice.

In several branches of the social sciences and humanities, surveys based on standardized questionnaires are a prominent research tool. While there are a variety of ways to analyze the data, some standard procedures have become established. When those surveys want to analyze differences in the answer patterns of different groups (e.g., countries, gender, age, ...), these procedures can only be carried out in a meaningful way if there is measurement invariance, i.e., the measured construct has psychometric equivalence across groups. As recently raised as an open problem by Sauerwein et al. (2021), new evaluation methods that work in the absence of measurement invariance are needed. This paper promotes an unsupervised learning-based approach to such research data by proposing a procedure that works in three phases: data preparation, clustering of questionnaires, and measuring similarity based on the obtained clustering and the properties of each group. We generate synthetic data in three data sets, which allows us to compare our approach with the PCA approach under measurement invariance and under violated measurement invariance. As a main result, we obtain that the approach provides a natural comparison between groups and a natural description of the response patterns of the groups. Moreover, it can be safely applied to a wide variety of data sets, even in the absence of measurement invariance. Finally, this approach allows us to translate (violations of) measurement invariance into a meaningful measure of similarity.