1
GPT-6 Astra 模型发布，OpenAI 总裁布罗克曼自信放话“欢迎来到 AGI 时代”
IT之家 9 月 5 日消息，据《商业内幕》报道，OpenAI 宣布，新推出的 Astra 模型已经达到 AI 行业最为追求的一个里程碑。当地时间周四，OpenAI 总裁格雷格 · 布罗克曼在宣布 Astra 模型推出的媒体电话会议结束时自信喊话：“ 欢迎来到 AGI 时代 。” OpenAI 将 …
2
Can Legal AI Know When It Is Wrong? And Do Students Know When It Is?
法律AI也会“迷之自信”？研究揭示大模型在司法判断中的过度自信陷阱，比你以为的更危险。
arXiv:2608.21089v1 Announce Type: new Abstract: Integrating Large Language Models (LLMs) into the Indian judiciary promises access to justice but intr…
3
When Confidence Fails: Overconfidence in LLMs under Uncertainty and Missing Clinical Information
当临床信息不完整时，大模型可能“迷之自信”，这项研究揭示风险并敲响医疗AI安全警钟。
arXiv:2608.09080v1 Announce Type: cross Abstract: Large Language Models (LLMs) have achieved strong performance in medical question answering and clin…
4
Confidently Deceptive: How Confidence Amplifies the Risk of LLM Deception
大模型越自信越容易说谎？这篇论文揭示置信度如何放大LLM的欺骗风险，对AI安全研究有重要启示。
arXiv:2607.20444v1 Announce Type: cross Abstract: Large language models (LLMs) can produce deceptive responses: outputs that mislead users in service …
5
Tutorial Completion Didn't Make Me Feel Like an Engineer
学完Linux、Docker、Nginx等全套教程，却依然怀疑自己是不是真工程师？这篇戳中无数自学者的共鸣痛点。
Over the past few months, I've completed hands-on learning across Linux, Docker, Docker Compose, Git, GitHub Actions, AWS EC2, Nginx, PostgreSQL, netw…
6
Calibration Drift Under Reasoning: How Chain-of-Thought Budgets Induce Overconfidence in Large Language Models
思维链推理预算越高，大模型反而越“迷之自信”？新研究揭示校准漂移现象与CABStop缓解方案
arXiv:2606.11211v1 Announce Type: cross Abstract: The ability of large language models (LLMs) to express calibrated uncertainty is important for safe …
7
RKSC: Reasoning-Aware KV Cache Sharing and Confident Early Exit for Multi-Step LLM Inference
多步LLM推理新突破：通过推理感知KV缓存共享与自信提前退出机制，大幅提升效率，已被ICML 2026 Workshop收录。
arXiv:2606.09937v1 Announce Type: cross Abstract: We introduce RKSC (Reasoning-Aware KV Cache Sharing), a training-free inference framework that elimi…
8
Large Language Models Are Overconfident in Their Own Responses
新研究表明指令微调大模型在自评时过度自信，揭示聊天模板对校准的关键影响。
arXiv:2606.03437v1 Announce Type: new Abstract: Prior work has shown that instruction-tuned large language models (LLMs) are less well calibrated than…
9
Claude 4.8炸场！部分能力超过Mythos，支持数百子智能体并行
Claude 4.8重磅升级：长任务自主执行、数百子智能体并行，错误率暴降10倍，代码缺陷报告减少75%
可以长时间执行任务，人类不用经常回来检查它的工作
10
王传福：比亚迪在 L2 阶段率先承担起 L3、L4 的责任是技术自信，更是为用户负责
IT之家 5 月 28 日消息，在今天的比亚迪智能化战略发布会上，比亚迪集团董事长兼总裁王传福发表演讲。 IT之家在发布会现场了解到，王传福宣布比亚迪继智能泊车安全兜底后， 再次率先承诺为城市领航安全兜底 1 年 。 王传福强调，比亚迪在 L2 阶段率先承担起 L3、L4 的责任是技术自信，更是为用…
11
Self-signals Driven Multi-LLM Debate for Efficient and Accurate Reasoning
多LLM辩论新范式：利用自信号驱动，在高效与准确推理间取得突破，节省算力并提升效果。
arXiv:2510.06843v2 Announce Type: replace-cross Abstract: Large Language Models (LLMs) have exhibited impressive capabilities across diverse applicati…
12
Understanding and Mitigating Premature Confidence for Better LLM Reasoning
最新研究揭示LLM长思维链中“过早自信”导致的逻辑缺口，并提出基于过程奖励模型的缓解策略，提升推理质量。
arXiv:2605.24396v1 Announce Type: new Abstract: Long chains of thought (CoT) from current language models frequently contain logical gaps and unjustif…
13
How do LLMs Compute Verbal Confidence
揭秘LLMs如何内部计算口头自信度：是即时响应还是自动生成？这项研究打开黑箱，深入模型推理过程。
arXiv:2603.17839v3 Announce Type: replace-cross Abstract: Verbal confidence -- prompting LLMs to state their confidence as a number or category -- is …