SayNext-Bench: Why Do LLMs Struggle with Next-Utterance Prediction?

• URL: http://arxiv.org/abs/2602.00327v1
• Date: Fri, 30 Jan 2026 21:27:45 GMT
• Title: SayNext-Bench: Why Do LLMs Struggle with Next-Utterance Prediction?
• Abstract summary: We show that even leading models surprisingly struggle to predict a human speaker's next utterance.<n>Instead, humans can readily anticipate forthcoming utterances based on multimodal cues, such as gestures, gaze, and emotional tone, from the context.<n>We develop a dual-route prediction MLLM, SayNext-Chat, that incorporates cognitively inspired design to emulate predictive processing in conversation.
• Score: 42.890703748349814
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: We explore the use of large language models (LLMs) for next-utterance prediction in human dialogue. Despite recent advances in LLMs demonstrating their ability to engage in natural conversations with users, we show that even leading models surprisingly struggle to predict a human speaker's next utterance. Instead, humans can readily anticipate forthcoming utterances based on multimodal cues, such as gestures, gaze, and emotional tone, from the context. To systematically examine whether LLMs can reproduce this ability, we propose SayNext-Bench, a benchmark that evaluates LLMs and Multimodal LLMs (MLLMs) on anticipating context-conditioned responses from multimodal cues spanning a variety of real-world scenarios. To support this benchmark, we build SayNext-PC, a novel large-scale dataset containing dialogues with rich multimodal cues. Building on this, we further develop a dual-route prediction MLLM, SayNext-Chat, that incorporates cognitively inspired design to emulate predictive processing in conversation. Experimental results demonstrate that our model outperforms state-of-the-art MLLMs in terms of lexical overlap, semantic similarity, and emotion consistency. Our results prove the feasibility of next-utterance prediction with LLMs from multimodal cues and emphasize the (i) indispensable role of multimodal cues and (ii) actively predictive processing as the foundation of natural human interaction, which is missing in current MLLMs. We hope that this exploration offers a new research entry toward more human-like, context-sensitive AI interaction for human-centered AI. Our benchmark and model can be accessed at https://saynext.github.io/.

Related papers

• PIVOTSBench: Evaluating Fine-Grained Interpersonal Relationship Reasoning in Multimodal Large Language Models [6.1391886258150805]
  We introduce PIVOTS, the first benchmark built from Social-IQ 2.0 and YouTube data to evaluate multimodal large language models (MLLMs)<n>PIVOTS includes auxiliary tasks that assess models' ability to identify and leverage the critical visual cues underlying such predictions.<n>We evaluate both proprietary and open-source MLLMs and conduct detailed ablation studies to analyze the effects of visual modalities and explicit social role information in conversational utterances.
  arXiv  Detail & Related papers  (2026-06-22T09:38:03Z)
• Evaluating Large Language Models Abilities for Addressee, Turn-change, and Next Speaker Prediction in Meetings [85.35443905857876]
  We investigate turn-taking in multimodal multi-party conversations using large language models (LLMs)<n>We construct an evaluation framework for three tasks: addressee detection, turn-change prediction, and next speaker prediction.<n>We compare supervised models trained for these tasks, text-based LLMs, multimodal LLMs (MM-LLMs), and human subjects.
  arXiv  Detail & Related papers  (2026-06-16T05:46:44Z)
• Can MLLMs Read the Room? A Multimodal Benchmark for Assessing Deception in Multi-Party Social Interactions [26.074938251210842]
  Despite advanced reasoning capabilities, state-of-the-art Multimodal Large Language Models (MLLMs) demonstrably lack a core component of human intelligence.<n>We introduce a new task, Multimodal Interactive Deception Assessment (MIDA)<n>We present a novel multimodal dataset providing synchronized video and text with verifiable ground-truth labels for every statement.
  arXiv  Detail & Related papers  (2025-11-20T10:44:21Z)
• SIV-Bench: A Video Benchmark for Social Interaction Understanding and Reasoning [53.16179295245888]
  We introduce SIV-Bench, a novel video benchmark for evaluating the capabilities of Multimodal Large Language Models (MLLMs) across Social Scene Understanding (SSU), Social State Reasoning (SSR), and Social Dynamics Prediction (SDP)<n>SIV-Bench features 2,792 video clips and 8,792 meticulously generated question-answer pairs derived from a human-LLM collaborative pipeline.<n>It also includes a dedicated setup for analyzing the impact of different textual cues-original on-screen text, added dialogue, or no text.
  arXiv  Detail & Related papers  (2025-06-05T05:51:35Z)
• Towards Anthropomorphic Conversational AI Part I: A Practical Framework [49.62013440962072]
  We introduce a multi- module framework designed to replicate the key aspects of human intelligence involved in conversations.<n>In the second stage of our approach, these conversational data, after filtering and labeling, can serve as training and testing data for reinforcement learning.
  arXiv  Detail & Related papers  (2025-02-28T03:18:39Z)
• Can LLMs Simulate Social Media Engagement? A Study on Action-Guided Response Generation [51.44040615856536]
  This paper analyzes large language models' ability to simulate social media engagement through action guided response generation.<n>We benchmark GPT-4o-mini, O1-mini, and DeepSeek-R1 in social media engagement simulation regarding a major societal event.
  arXiv  Detail & Related papers  (2025-02-17T17:43:08Z)
• OmniBench: Towards The Future of Universal Omni-Language Models [63.16606414452612]
  We introduce OmniBench, a novel benchmark designed to evaluate models' ability to recognize, interpret, and reason across visual, acoustic, and textual inputs simultaneously.<n>Our evaluation reveals that open-source OLMs show significant limitations in instruction-following and reasoning in tri-modal contexts.<n>We advocate for developing more robust tri-modal integration techniques and training strategies to enhance OLM performance.
  arXiv  Detail & Related papers  (2024-09-23T17:59:05Z)
• Traj-LLM: A New Exploration for Empowering Trajectory Prediction with Pre-trained Large Language Models [12.687494201105066]
  This paper proposes Traj-LLM, the first to investigate the potential of using Large Language Models (LLMs) to generate future motion from agents' past/observed trajectories and scene semantics. LLMs' powerful comprehension abilities capture a spectrum of high-level scene knowledge and interactive information. Emulating the human-like lane focus cognitive function, we introduce lane-aware probabilistic learning powered by the pioneering Mamba module.
  arXiv  Detail & Related papers  (2024-05-08T09:28:04Z)
• Benchmarking Sequential Visual Input Reasoning and Prediction in Multimodal Large Language Models [21.438427686724932]
  We introduce a novel benchmark that assesses the predictive reasoning capabilities of MLLMs across diverse scenarios. Our benchmark targets three important domains: abstract pattern reasoning, human activity prediction, and physical interaction prediction. Empirical experiments confirm the soundness of the proposed benchmark and evaluation methods.
  arXiv  Detail & Related papers  (2023-10-20T13:14:38Z)
• Let Models Speak Ciphers: Multiagent Debate through Embeddings [84.20336971784495]
  We introduce CIPHER (Communicative Inter-Model Protocol Through Embedding Representation) to address this issue. By deviating from natural language, CIPHER offers an advantage of encoding a broader spectrum of information without any modification to the model weights. This showcases the superiority and robustness of embeddings as an alternative "language" for communication among LLMs.
  arXiv  Detail & Related papers  (2023-10-10T03:06:38Z)
• Sight Beyond Text: Multi-Modal Training Enhances LLMs in Truthfulness and Ethics [32.123919380959485]
  Multi-modal large language models (MLLMs) are trained based on large language models (LLM) While they excel in multi-modal tasks, the pure NLP abilities of MLLMs are often underestimated and left untested. We show that visual instruction tuning, a prevailing strategy for transitioning LLMs into MLLMs, unexpectedly and interestingly helps models attain both improved truthfulness and ethical alignment.
  arXiv  Detail & Related papers  (2023-09-13T17:57:21Z)
• Multimodality and Attention Increase Alignment in Natural Language Prediction Between Humans and Computational Models [0.8139163264824348]
  Humans are known to use salient multimodal features, such as visual cues, to facilitate the processing of upcoming words. multimodal computational models can integrate visual and linguistic data using a visual attention mechanism to assign next-word probabilities. We show that predictability estimates from humans aligned more closely with scores generated from multimodal models vs. their unimodal counterparts.
  arXiv  Detail & Related papers  (2023-08-11T09:30:07Z)
• A Survey on Multimodal Large Language Models [71.63375558033364]
  Multimodal Large Language Model (MLLM) represented by GPT-4V has been a new rising research hotspot.<n>This paper aims to trace and summarize the recent progress of MLLMs.
  arXiv  Detail & Related papers  (2023-06-23T15:21:52Z)
• Chat with the Environment: Interactive Multimodal Perception Using Large Language Models [19.623070762485494]
  Large Language Models (LLMs) have shown remarkable reasoning ability in few-shot robotic planning. Our study demonstrates that LLMs can provide high-level planning and reasoning skills and control interactive robot behavior in a multimodal environment.
  arXiv  Detail & Related papers  (2023-03-14T23:01:27Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.