Humor is a complex form of communication that remains challenging for machines. Despite its broadness, most existing research on computational humor traditionally focused on modeling one specific type of humor. In this work, we wish to understand whether competence on specific humor tasks confers any ability to transfer to novel, unseen types; in other words, is this fragmentation inevitable? This question is especially timely as new humor types continuously emerge in online contexts (e.g., memes, anti-humor, AI fails). If LLMs are to keep up with this evolving landscape, they must be able to capture deeper, transferable mechanisms. To investigate this, we conduct a series of transfer learning experiments across four datasets, representing different humor tasks. We explore varied diversity settings (varying between 1-3 datasets in training, testing on a novel one). Experiments show that models are capable of some transfer, reaching up to 75% accuracy on binary unseen datasets; training on diverse sources improves transferability (1.88-4.05%) with minimal-to-no drop in in-domain performance. Somewhat surprisingly, the one dataset (Dad Jokes) emerges as the best enabler of transfer, but the hardest one to transfer to. We release data and code.

Stand-up comedy, and humor in general, are often studied through their verbal content. Yet live performance relies just as much on embodied presence and audience feedback. We introduce TIC-TALK, a multimodal resource with 5,400+ temporally aligned topic segments capturing language, gesture, and audience response across 90 professionally filmed stand-up comedy specials (2015–2024). The pipeline combines BERTopic for 60 s thematic segmentation with dense sentence embeddings, Whisper-AT for 0.8 s laughter detection, a fine-tuned YOLOv8-cls shot classifier, and YOLOv8s-pose for raw keypoint extraction at 1 fps. Raw 17-joint skeletal coordinates are retained without prior clustering, enabling the computation of continuous kinematic signals—arm spread, kinetic energy, and trunk lean—that serve as proxies for performance dynamics. All streams are aligned by hierarchical temporal containment without resampling, and each topic segment stores its sentence-BERT embedding for downstream similarity and clustering tasks. As a concrete use case, we study laughter dynamics across 24 thematic topics: kinetic energy negatively predicts audience laughter rate (r=−0.75,N= 24), consistent with a stillness-before-punchline pattern; personal and bodily content elicits more laughter than geopolitical themes; and shot close-up proportion correlates positively with laughter (r= +0.28), consistent with reactive montage.

Arabic humor provides a challenging diagnostic test for large language models because interpreting jokes often requires pragmatic inference, sociolinguistic awareness, and culturally grounded knowledge that standard NLP benchmarks do not evaluate. Arabic is particularly suitable for probing these abilities given its diglossic structure and dialect diversity, where humor frequently arises from register contrast, dialect-specific vocabulary, and shared cultural references. We propose a three-layer taxonomy of Arabic humor mechanisms covering pragmatic, semantic, and sociolinguistic phenomena, illustrated through thirteen curated examples spanning Egyptian, Levantine, Gulf, Tunisian, and Iraqi Arabic. Building on this taxonomy, we introduce a diagnostic evaluation framework using contrastive minimal pairs, a multi-dimensional scoring rubric, and a cultural presupposition ontology. A small proof-of-concept probing study with GPT-4o, Gemini 2.0 Flash, and Claude Sonnet 4.5 reveals recurring failure patterns in sarcasm interpretation, register contrast reasoning, dialectal vocabulary coverage, and cultural grounding. We position this work as a diagnostic framework and pilot, not a mature benchmark, and outline a path toward larger annotated resources.

Humor is one of the most culturally embedded and socially significant dimensions of human communication, yet it remains largely unexplored as a dimension of Large Language Model (LLM) alignment. In this study, five frontier language models play the same Cards Against Humanity games (CAH) as human players. The models select the funniest response from a slate of ten candidate cards across 9,894 rounds. While all models exceed the random baseline, alignment with human preference remains modest. More striking is that models agree with each other substantially more often than they agree with humans. We show that this preference is partly explained by systematic position biases and content preferences, raising the question whether LLM humor judgment reflects genuine preference or structural artifacts of inference and alignment.

We present exploratory experiments in the comedic roasting capabilities of GPT4o. Specifically, @ComedyCentral roasts were scraped to design a survey in which participants blindly evaluated snippets of human and AI roasts, and had to predict the author (AI/human) in a second round of reviewing. The results show that there is no significant difference in how the barbs in human- and AI-generated roasts are rated. Further, a qualitative analysis showed that although the model utilizes specific recurrent phrases to imitate the style of human comedians, both generative LLM detectors and humans performed suboptimally in predicting the true author of the roasts.

This paper studies joke detection in short text, focusing only on jokes triggered by lexical ambiguity. Following Attardo and Raskin, we treat these jokes as cases where humor arises from a script opposition activated through a logical mechanism such as homography or homophony. Our framework combines contextuals emantic analysis for homographs with phoneme-level similarity for homophones and near-homophones, using CMUdict, weighted Levenshtein distance, and prompt-based reasoning to recover ambiguities that are not visible in spelling alone. Results show that explicit phonetic modeling improves detection of sound-based puns.

We investigate whether scaling model parameters improves humor generation through a controlled ablation study. Using five Qwen3 variants (8B–235B, dense and MoE), we generate jokes across 50 themes. Beyond evaluating humor scaling, this work serves as an empirical study into the nature of LLM versus human evaluations on highly subjective creative tasks. While an automated judge yields a perfect monotonic ranking between parameter count and win rate, human annotators find no significant aggregate difference in humor quality. Restricting to themes where annotators agree reveals a significant preference for the largest model (p = 0.039), suggesting scaling effects exist but are masked by a "quality floor." Crucially, our analysis of bias characteristics shows that the automated judge exhibits severe positional and length biases compared to human evaluators, further suggesting that LLMs may systematically distort quality differences on subjective tasks.

This study validates automated, corpus-based methods for quantifying joke originality using “topic handles” — key nouns or noun phrases capturing a joke’s script opposition and logical mechanism (per the General Theory of Verbal Humor). Using a reference corpus of one million jokes in English from Reddit, we compute Pointwise Mutual Information (PMI) in three variants (raw co-occurrence, semantic-cluster smoothing, and word-decomposition) and two embedding-based measures (handle-level conceptual distance and full-text corpus novelty via Sentence-BERT). We evaluate these measures on 400 LLM-generated jokes (200 each from GPT-4o and GPT-5.4) and 80 jokes from the Witscript-powered JEST benchmark, rated by three professional comedians for originality and funniness. Corpus novelty and concept distance between the most semantically distant handle pair both correlated significantly with human originality ratings (𝜌 = .37); PMI-based measures showed weaker but significant associations (𝜌 = .23–.25) on the most original handle pair. A Lasso-based composite of the three strongest predictors achieved 𝜌 = .40 (cross-validated), capturing 82% of the theoretically predictable variance given inter-rater agreement. These results demonstrate that handle-based PMI and semantic novelty metrics offer practical, quantitative tools for assessing originality in AI-generated humor, advancing objective evaluation of computational creativity.