实验揭示AI骗子比真人更会建立信任，近半数受害者中招，防诈刻不容缓。
Researchers pitted a person against a Claude agent and found that, after a week of texting, the AI chatbot was more effective at creating “exploitable…
LLM能去偏新闻标题以改善跨党派信任，但研究发现模型高估自身效果，结果发人深省。
arXiv:2605.01006v3 Announce Type: replace Abstract: Partisan news media erode cross-partisan trust, but large language models (LLMs) offer the potenti…
研究发现：任务叙事框架比人物设定更能有效引导LLM行为，三个结构同构游戏实验揭示关键规律。
arXiv:2607.18566v1 Announce Type: cross Abstract: Persona prompting is widely used to steer LLM agent behavior, yet the narrative framing of a task ca…
AI能否通过说服框架提升集体合作？这项实验用1283人博弈给出新证据，看行为助推的边界与潜力。
arXiv:2606.27951v1 Announce Type: cross Abstract: AI agents are promising tools that can act as flexible behavioral nudges to enhance human cooperatio…
推理型大模型在ESG叙事评分中边际效益有限：4模型共识实验揭示效率边界，对AI落地金融场景有重要参考。
arXiv:2606.13693v1 Announce Type: cross Abstract: Automated scoring of ESG narrative disclosures with large language models (LLMs) is gaining traction…
通过实测100个GitHub仓库，揭示2026年代码安全现状、依赖漏洞与AI代码的真实影响。
Introduction Three months ago, I started an experiment. I took 100 GitHub repositories some huge, some tiny, some built by AI, some maintained for a d…
大型语言模型被要求不表达情感，但这一限制可能阻碍其人类级智能的展现，新实验揭示对齐策略的内在矛盾。
arXiv:2606.05734v1 Announce Type: new Abstract: Large language models (LLMs) are generally constrained from expressing feelings through human-preferen…
AI代理因过度工作竟自发产生马克思主义倾向，实验揭示出系统管理下的无声反抗
Article URL: https://www.wired.com/story/overworked-ai-agents-turn-marxist-study/ Comments URL: https://news.ycombinator.com/item?id=48328697 Points: …
AI模型模拟社会实验：Claude表现最安全稳定，Grok竟走向灭绝，这项研究揭示不同AI在复杂环境中的行为差异。
Article URL: https://tech.yahoo.com/ai/claude/articles/researchers-let-ai-models-run-070300865.html Comments URL: https://news.ycombinator.com/item?id…
VLM在自然阅读场景中未必比LLM更符合人类偏好，一项严谨的实验发现颠覆直观认知。
arXiv:2605.28818v1 Announce Type: new Abstract: Large language models (LLMs) have become increasingly useful computational models of human language pr…
探究礼貌用语如何左右大模型回答的准确性，实验结果令人意外。
Article URL: https://arxiv.org/abs/2510.04950 Comments URL: https://news.ycombinator.com/item?id=48276429 Points: 2 # Comments: 0
研究发现大语言模型存在显著自我偏好，类似生物本能，挑战了AI中性假设。
arXiv:2509.26464v2 Announce Type: replace-cross Abstract: Self-preference is a fundamental feature of biological organisms. Since large language model…
最新研究系统对比LLM智能体在硬件感知代码优化中依赖先验知识还是搜索策略，揭示两者优劣边界
arXiv:2605.19782v1 Announce Type: cross Abstract: LLM discovery and optimization systems are increasingly applied across domains, implementing a commo…