Generative AI offers scalable engagement with digital cultural heritage, yet current workflows rely on complex prompts, creating an interaction bottleneck for non-expert audiences. To address this gap, we introduce MoodCurator, a web-based design probe enabling low-friction, implicit affective steering for curatorial storytelling. The system replaces prompt engineering with a three-channel loop: audiences (1) select a colour palette to suggest narrative tone; (2) curate artworks to ground the narrative in visual content; and (3) choose an interpretive voice to constrain rhetorical stance. A mixed-methods study (N = 64) demonstrated encouraging patterns of perceived usability and user agency. A follow-up think-aloud study (N = 10) surfaced recurring cases of interpretive mismatch in this English-language deployment. Contributions include: (1) a legible steering paradigm for AI-supported cultural interpretation; (2) exploratory empirical evidence on usability, agency, and narrative resonance; and (3) preliminary design implications for more explicit and contestable framing controls in future systems.

Recent breakthroughs in Large Language Models (LLMs) have positioned them as a promising paradigm for agents, with long-term planning and decision-making emerging as core general-purpose capabilities for adapting to diverse scenarios and tasks. Real-time strategy (RTS) games serve as an ideal testbed for evaluating these two capabilities, as their inherent gameplay requires both macro-level strategic planning and micro-level tactical adaptation and action execution. Existing RTS game-based environments either suffer from relatively high computational demands or lack support for textual observations, which has constrained the use of RTS games for LLM evaluation. Motivated by this, we present TowerMind, a novel environment grounded in the tower defense (TD) subgenre of RTS games. TowerMind preserves the key evaluation strengths of RTS games for assessing LLMs, while featuring low computational demands and a multimodal observation space, including pixel-based, textual, and structured game-state representations. In addition, TowerMind supports the evaluation of model hallucination and provides a high degree of customizability. We design five benchmark levels to evaluate several widely used LLMs under different multimodal input settings. The results reveal a clear performance gap between LLMs and human experts across both capability and hallucination dimensions. The experiments further highlight key limitations in LLM behavior, such as inadequate planning validation, a lack of multifinality in decision-making, and inefficient action use. We also evaluate two classic reinforcement learning algorithms: Ape-X DQN and PPO. By offering a lightweight and multimodal design, TowerMind complements the existing RTS game-based environment landscape and introduces a new benchmark for the AI agent field.

Dawei Wang, Di Zhao, Xinyuan Liu, Marci Chi Ma, Xiaoyang Liu, Chengming Zhou, Gary Ushaw, Richard Davison. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.