arXiv cs.CL· David Melhart, Matthew Barthet, Georgios N. Yannakakis·· 12 小时前AI 评分29
大语言模型能否捕捉电子游戏参与度？LLM 预测游戏投入感仍逊于人类
Can Large Language Models Capture Video Game Engagement?
AI 导读
研究首次系统评估主流 LLM 在多模态提示下预测视频连续情感标注的能力，基于 GameVibe 语料库中 20 款第一人称射击游戏的 80 分钟标注录像，开展超 4,800 次实验，考察模型架构、规模、输入模态、提示策略与真值处理方法的影响。结果显示 LLM 虽能超越传统机器学习基线，但整体仍落后于人类提供的连续体验标注，且在不同游戏间表现波动明显。
正文 · 原文
Abstract:Can out-of-the-box pretrained Large Language Models (LLMs) detect human affect successfully when observing a video? To address this question, for the first time, we evaluate comprehensively the capacity of popular LLMs for successfully predicting continuous affect annotations of videos when prompted by a sequence of text and video frames in a multimodal fashion. In this paper, we test LLMs' ability to correctly label changes of in-game engagement in 80 minutes of annotated videogame footage from 20 first-person shooter games of the GameVibe corpus. We run over 4,800 experiments to investigate the impact of LLM architecture, model size, input modality, prompting strategy, and ground truth processing method on engagement prediction. Our findings suggest that while LLMs rightfully claim human-like performance across multiple domains and able to outperform traditional machine learning baselines, they generally fall behind continuous experience annotations provided by humans. We examine some of the underlying causes for a fluctuating performance across games, highlight the cases where LLMs exceed expectations, and draw a roadmap for the further exploration of automated emotion labelling via LLMs.
| Comments: | Version accepted for publication at the IEEE Transactions on Affective Computing |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC) |
| Cite as: | arXiv:2502.04379 [cs.CV] |
| (or arXiv:2502.04379v3 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2502.04379 arXiv-issued DOI via DataCite |
Submission history
From: David Melhart PhD [view email]
[v1]
Wed, 5 Feb 2025 17:14:47 UTC (16,198 KB)
[v2]
Thu, 29 Jan 2026 08:44:15 UTC (13,775 KB)
[v3]
Mon, 2 Nov 2026 14:13:19 UTC (16,032 KB)
来源：arXiv cs.CL · arxiv.org