This paper defines the "Plausibility Trap," a phenomenon where individuals over-rely on expensive probabilistic Large Language Models (LLMs) for simple deterministic tasks, leading to significant resource waste and risks like algorithmic sycophancy. It introduces a framework for proper tool selection and advocates for a curriculum shift in digital literacy.

This paper investigates the prevalence and mechanisms of social bias in Large Vision-Language Model (LVLM)-based text-to-image (T2I) models, revealing that system prompts are a key driver. It introduces FairPro, a training-free meta-prompting framework that enables LVLMs to dynamically generate fairness-aware system prompts, significantly reducing bias while maintaining image quality.