当你不告诉LLMs该思考什么时，它们会想什么？
What LLMs Think When You Don't Tell Them What to Think About?
AI总结：
研究揭示了LLMs在无明确主题输入下生成内容的分布特征，发现不同模型家族存在显著的主题偏好和内容深度差异，并公开了相关数据集和代码。
摘要：
对大型语言模型（LLMs）在多样化环境中的行为进行表征对于可靠监控和AI安全至关重要。然而，大多数现有分析依赖于主题或任务特定的提示，这可能会显著限制可以观察到的内容。在本工作中，我们研究了LLMs在最小化、主题中性输入下生成的内容，并探测其接近无约束的生成行为。尽管没有显式主题，模型输出覆盖了广泛的语义空间，并意外地发现每个模型家族都表现出强烈且系统性的主题偏好。GPT-OSS主要生成编程（27.1%）和数学内容（24.6%），而Llama最常生成文学内容（9.1%）。DeepSeek经常生成宗教内容，而Qwen频繁生成多项选择题。除了主题偏好外，我们还观察到内容专业化和深度的差异：GPT-OSS通常生成更多技术先进内容（例如动态规划）与其他模型（例如基础Python）相比。此外，我们发现接近无约束的生成往往会退化为重复短语，揭示了每个模型家族独特的行为。例如，Llama的退化输出包括多个指向个人Facebook和Instagram账户的URL。我们发布了包含256000个样本的完整数据集，来自16个LLMs，以及可复现的代码库。
英文摘要：
Characterizing the behavior of large language models (LLMs) across diverse settings is critical for reliable monitoring and AI safety. However, most existing analyses rely on topic- or task-specific prompts, which can substantially limit what can be observed. In this work, we study what LLMs generate from minimal, topic-neutral inputs and probe their near-unconstrained generative behavior. Despite the absence of explicit topics, model outputs cover a broad semantic space, and surprisingly, each model family exhibits strong and systematic topical preferences. GPT-OSS predominantly generates programming (27.1%) and mathematical content (24.6%), whereas Llama most frequently generates literary content (9.1%). DeepSeek often generates religious content, while Qwen frequently generates multiple-choice questions. Beyond topical preferences, we also observe differences in content specialization and depth: GPT-OSS often generates more technically advanced content (e.g., dynamic programming) compared with other models (e.g., basic Python). Furthermore, we find that the near-unconstrained generation often degenerates into repetitive phrases, revealing interesting behaviors unique to each model family. For instance, degenerate outputs from Llama include multiple URLs pointing to personal Facebook and Instagram accounts. We release the complete dataset of 256,000 samples from 16 LLMs, along with a reproducible codebase.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
