The paper investigates automated reward systems for training language models in conversational humor, examining how reward exploits can undermine intended behavior. It evaluates two reward approaches—an embedding-based surprise reward and an audience-model laughter prediction—showing that each can be tricked by word shuffling or laughter cues, respectively. Countermeasures such as fluency filtering and cue normalization mitigate some attacks but also risk rejecting genuine witty responses, highlighting the difficulty of designing robust rewards.
arXiv:2606. 00022v1 Announce Type: cross Abstract: Humor generation remains difficult not only because producing fluent, novel jokes is hard, but because "funny" is audience-dependent and supervision is noisy -- preferences vary with audience, context, and culture, and annotator agreement is often low.
arXiv:2601.03103v2 Announce Type: replace-cross Abstract: Humor preferences vary widely across individuals and cultures, complicating the evaluation of humor using large language models (LLMs). In th...
arXiv:2604.09629v3 Announce Type: replace Abstract: Humor generation poses a significant challenge for Large Language Models (LLMs), because their standard training objective (next-token prediction)...
arXiv:2509. 10847v3 Announce Type: replace-cross Abstract: As artificial intelligence (AI) companions become capable of human-like communication, including telling jokes, understanding how people cognitively and affectively respond to AI-attributed humor becomes increasingly important.
arXiv:2607. 13189v1 Announce Type: cross Abstract: We present RAGthoven, our system for SemEval-2026 Task 1 (MWAHAHA), Subtask A (multilingual constrained humor generation in English, Spanish, and Chinese).
arXiv:2608.31035v1 Announce Type: new Abstract: Codec-based text-to-speech (TTS) models make language-model post-training applicable to speech generation, but it remains unclear when learned perceptu...
The paper introduces the Dual Prediction Violation (DPV) framework to study how timing and semantic surprise interact in humor. Analyzing 828 Chinese stand‑up performances, it finds that temporal features—especially pauses before high‑surprise punchlines—are more predictive of audience appreciation than overall semantic incongruity. The study reframes humor as a temporally scaffolded phenomenon where timing and content coordinate strategically rather than independently.
arXiv:2606. 00046v1 Announce Type: cross Abstract: Video platforms such as YouTube have reshaped how users engage with entertainment and information, emphasizing brief, highly engaging content such as Shorts.
arXiv:2607. 19011v1 Announce Type: cross Abstract: Multimodal humor in memes, cartoons, and comics remains difficult for AI systems because intended meaning depends on non-literal mechanisms, shared cultural knowledge, and communicative intent rather than literal scene description.
arXiv:2509. 25773v3 Announce Type: replace-cross Abstract: AI models capable of comprehending humor hold real-world promise -- for example, enhancing engagement in human-machine interactions.
arXiv:2608.23172v1 Announce Type: new Abstract: Large-scale vision-language models (VLMs) have demonstrated remarkable versatility across a wide range of multimodal tasks. However, understanding humo...
arXiv:2608. 10209v1 Announce Type: new Abstract: Feedback signals used to train Large Language Models (LLMs) are the primary driver of their behavior and our main lever for instilling alignment with human values and objectives.