1
联合国 AI 专家反对 AI 末日论，呼吁理性讨论
IT之家 9 月 22 日消息，联合国下属人工智能委员会的多名专家于当地时间周一发出提醒，反对业内部分从业者针对 AI 风险所发表的世界末日式言论。 独立机构国际 AI 科学专家组（International Scientific Panel on AI）成员、谷歌 DeepMind 高管若埃尔 ·…
2
索尼辩称“理性消费者”都知道，在 PS Store 买了数字游戏并不等于取得所有权
IT之家 9 月 1 日消息，据外媒 Wccftech 今天（1 日）报道，索尼强烈反驳了一项针对 PlayStation 数字游戏许可条款的拟议集体诉讼。公司在最近提交的法庭文件中主张，一个 正常理解数字商品交易方式的“理性消费者” 应该知道，在 PlayStation Store 买下一款数字游…
3
LLM Inference Under Bursty Workload Distribution: Modifying the WAIT Algorithm
应对突发推理负载，改进WAIT调度算法，提升LLM服务稳定性与吞吐量。
arXiv:2608.06135v1 Announce Type: new Abstract: Large Language Models (LLMs) such as ChatGPT and Claude are widely used for information retrieval and …
4
Trust Is Gone: AI Safety Needs Individuals
机构信任崩塌后，AI安全责任如何落到每个人身上？一场关乎人类未来的个体觉醒论述。
Article URL: https://www.lesswrong.com/posts/cm3Ea9tBHpjj9zjji/trust-is-gone-ai-safety-needs-individuals Comments URL: https://news.ycombinator.com/it…
5
Running a 35B LLM at 128K Context, Full Speed, on €870 of Used Hardware
870欧元二手硬件全速跑35B大模型，128K上下文本地部署的极限性价比方案。
Article URL: https://medium.com/ai-advances/running-a-35b-llm-at-128k-context-full-speed-on-870-of-used-hardware-no-cloud-required-c4f7629810b8 Commen…
6
Nothing Is Easy When You're an LLM: The Flat Latency Problem
大模型推理的“扁平延迟”困境，为何优化如此棘手？
Article URL: https://aidoses.substack.com/p/nothing-is-easy-when-youre-an-llm Comments URL: https://news.ycombinator.com/item?id=49125443 Points: 1 # …
7
工信部召开重点车企座谈会，强调抵制非理性竞争、加强自动驾驶安全评估
工信部出手规范汽车行业，聚焦自动驾驶安全评估与质量监管，释放行业整治信号
IT之家 7 月 17 日消息，今日，工业和信息化部装备工业一司组织召开重点汽车生产企业座谈会，部署进一步规范汽车产业竞争秩序、提升汽车产品生产一致性和质量安全水平、开展汽车产品安全风险隐患排查和监督检查等重点工作。 会议要求，汽车生产企业要 坚决抵制非理性竞争 ，加强产品测试验证与安全评估，守牢产…
8
报告称AI收入已达临界点，开始证明数据中心投资的经济合理性
AI收入突破临界点，数据中心投资迎来经济合理性的关键证明。
研究机构Exponential View的报告显示，来自于人工智能（AI）的收入已达临界点，表明科技公司在这一领域投入数千亿美元在经济上可能具有可持续性。报告显示，超大规模云服务商和新兴云服务商的全球AI销售收入达到250亿美元，连续第二个季度超过该行业与数据中心和芯片投资相关的估计折旧成本210亿…
9
More Convincing, Not More Correct: Self-Play Reward Hacking of Reference-Free LLM Judges
自博弈评判者打分的是“像不像”而非“对不对”，参考无关LLM裁判存在结构性缺陷。
arXiv:2607.05904v1 Announce Type: new Abstract: Training a language model against its own reference-free judgments (the premise of self-rewarding, sel…
10
Please stop the AI confidence theater
AI宣传的“确定性”正在毒害理性思考，拆穿AI乐观情绪背后的认知陷阱，值得冷静一读。
Article URL: https://www.elenaverna.com/p/please-stop-the-ai-confidence-theater Comments URL: https://news.ycombinator.com/item?id=48774414 Points: 97…
11
A Practice Auditing Framework for Large Language Model Use: Collective Empiricism, Pseudo-Rational Cognition, and Governance of AI-Generated Content
一套针对大模型使用的实践审计框架，直击AI生成内容治理盲区，为理性评估与风险管控提供新思路。
arXiv:2607.01248v1 Announce Type: cross Abstract: Large language models are increasingly used for knowledge acquisition, code generation, academic wri…
12
Predicting LLM Reasoning Performance with Small Proxy Model
用小代理模型精准预测LLM推理性能，省钱省算力的高效新方案。
arXiv:2509.21013v4 Announce Type: replace-cross Abstract: Given the prohibitive cost of pre-training large language models, it is essential to leverag…
13
Togetherness: How co-operation built the world
一针见血地解读共生与合作如何推动生命演化进程，提供全新的科学认知视角，内容丰富且启发思考。
Nature, Published online: 01 July 2026; doi:10.1038/d41586-026-02067-6 Science journalist Rowan Hooper joins us to talk about symbiosis and the import…
14
Why AI Hasn't Cured Anything yet [video]
AI在医疗领域高歌猛进，却至今没治好一种病：这期视频冷静拆解了背后的现实与瓶颈。
Article URL: https://www.youtube.com/watch?v=n-hWHV2ZKOA Comments URL: https://news.ycombinator.com/item?id=48738356 Points: 1 # Comments: 0
15
Make AI Boring Again
AI被吹得神乎其神？这位资深博主呼吁让AI回归平凡，当它真正“无聊”时才是好技术。
Article URL: https://charitydotwtf.substack.com/p/make-ai-boring-again Comments URL: https://news.ycombinator.com/item?id=48669707 Points: 3 # Comment…
16
Arguments that Alter Minds: LLM Rationales Sway Human (and LLM) Notions of Plausibility
LLM生成的论证竟能同时左右人类和模型的合理性判断，揭示人机交互中微妙的说服力效应。
arXiv:2510.08091v3 Announce Type: replace-cross Abstract: We investigate the degree to which human (and LLM) plausibility judgments of multiple-choice…
17
In LLM Reasoning, there is Irrationality on top of Value Misalignment
即使对齐训练完美，大模型推理仍会偏离目标价值，文中用数学形式化定义了这一风险缺口。
arXiv:2606.20624v1 Announce Type: new Abstract: Significant progress has been made in aligning LLMs with target value functions. We argue that, even w…
18
GEOPHYS: The Geometry of Physical Plausibility
最新arXiv论文提出GEOPHYS框架，用几何方法量化物理合理性，为AI生成内容提供可解释性评估。
arXiv:2606.20707v1 Announce Type: cross Abstract: While humans can identify physically implausible events within milliseconds, machine learning approa…
19
Show HN: Local LLM Hardware Calculator
想跑本地大模型又怕带不动？这个计算器根据有效参数量、MoE 和 KV-cache 估算硬件需求，比单纯看显存靠谱得多。
Article URL: https://vettedconsumer.com/can-i-run-it/ Comments URL: https://news.ycombinator.com/item?id=48618065 Points: 1 # Comments: 0
20
Some Thoughts on AI Safety
不跟风不贩卖焦虑，用工程视角拆解AI安全议题，理性又清醒。
Article URL: https://stevekinney.com/writing/thoughts-on-ai-safety Comments URL: https://news.ycombinator.com/item?id=48601307 Points: 1 # Comments: 0