AAI Search
AI 资讯日报 · 2026-08-16

2026-08-16 · AI 资讯日报

今日新收录 56 条公开资讯，按模型 / 产品 / 行业 / 论文 / 观点 自动归类汇编（非 AI 生成，点击可溯源原文）。

今日精选6 条

  1. 1.
    行业动态The AI Credit Resale Economy— Hacker News
  2. 2.
    行业动态Claude: System Prompts— Hacker News
  3. 3.
    模型发布/更新than two thirds of the zeros of the riemann zeta function are simple and on the critical line— Anthropic

    than two thirds of the zeros of the riemann zeta function are simple and on the critical line www-cdn.anthropic.com

  4. 4.
    行业动态Optima tackles AI benchmarking's biggest flaw by letting users test models against their own data— The Decoder

    Artificial Analysis has launched Optima, a platform that lets users build custom AI benchmarks from their own data and workflows. Models can be compared not just on quality but als…

  5. 5.
    论文研究One in five US workers now delegates tasks to AI instead of colleagues, survey finds— The Decoder

    A representative survey by Epoch AI found that 20 percent of employed Americans hand off at least one task to AI that a human used to do. Generally, they accept AI output with litt…

  6. 6.
    行业动态Top mathematicians say LLMs are strong calculators but poor creative thinkers— The Decoder

    Two renowned mathematicians, Timothy Gowers and Peter Sarnak, say large language models are good at combining known methods but lack the intuition for genuinely new mathematical id…

模型发布/更新6 条

  1. 1.
    than two thirds of the zeros of the riemann zeta function are simple and on the critical line— Anthropic

    than two thirds of the zeros of the riemann zeta function are simple and on the critical line www-cdn.anthropic.com

  2. 2.
    Z.ai Ships GLM-5.3 Without Retraining the Base Model: Better at Complex Coding and Long-Horizon Tasks— MarkTechPost

    Z.ai released GLM-5.3 on August 14, 2026. The model reuses the 743B GLM-5.2 base unchanged. Every reported gain comes from scaled post-training: more long-horizon task environments…

  3. 3.
    SpaceXAI Releases Grok 4.6: A 500K-Context Frontier Model Tuned for Long-Running Agents, Coding, and Knowledge Work— MarkTechPost

    SpaceXAI released Grok 4.6 on August 12, 2026 — a post-training upgrade over Grok 4.5, not a larger base model. It ties GPT-5.6 Sol Max at 61 on the Artificial Analysis Intelligenc…

  4. 4.
    Google AI Just Released Gemini 3.7 Flash: A Coding and Agent Model at $0.75/1M Input Tokens— MarkTechPost

    Google has released Gemini 3.7 Flash, a refinement of Gemini 3.6 Flash with algorithmic improvements to its reasoning core. It handles text, images, audio, and video across a 1M-to…

  5. 5.
    NVIDIA AI Releases Nemotron 3.5 Lightning: A 30B Open MoE with 3B Active Parameters, and NeMo Switchyard Model Router— MarkTechPost

    NVIDIA's open 30B MoE targets the agent execution layer, with Switchyard routing each step to the cheapest capable model. The post NVIDIA AI Releases Nemotron 3.5 Lightning: A 30B…

  6. 6.
    Qwen 3.8 27B is excellent, but it defaults to wildly overthinking things— Simon Willison

    AI 导读 · 小参数旗舰模型兼顾性能与效率，但过度思考问题暴露推理成本隐忧。

产品发布/更新6 条

  1. 7.
    zouyuxuan122/Deepseek-Harness-EAC— GitHub

    DeepSeek Harness Desktop (dsh-desktop). EAC: Embracing All Creation (揽尽万象). Bundled Node.js runtime with full dsh-CLI kernel, one-click startup, 10 built-in UI…

  2. 8.
    Dyna Robotics Introduces Dyna-2: A World-Action Model Pre-Trained on 1 Million Hours of Human Video— MarkTechPost

    Dyna Robotics has released Dyna-2, a world-action model pre-trained on more than one million hours of egocentric human video. The technical report establishes three results: a scal…

  3. 9.
    斯坦福调查显示：84% 中国受访者对 AI 感到兴奋，美国比例仅为 38%— IT之家

    AI 导读 · 中美AI态度悬殊揭示技术落地与社会信任差异，反映产业生态和文化认同的深层博弈。

  4. 10.
    中消协：近期 AI 客服相关投诉显著增多— IT之家

    AI 导读 · AI客服成“拦路虎”，用户维权通道被堵，服务体验亟待正视。

  5. 11.
    杭州95后小伙，5亿买走硅谷城堡— 量子位

    AI 导读 · 中国资本收购硅谷地标，青年创业者出海新信号。

  6. 12.
    “B站教AI”爆火后，北航90后副教授何静回应一切— 量子位

    AI 导读 · 学术新星直面流量，展现技术传播与公众对话的鲜活样本。

行业动态6 条

  1. 13.
    The AI Credit Resale Economy— Hacker News

    AI 导读 · AI信用倒卖暗藏套利与合规风险，预示新灰色市场崛起。

  2. 14.
    Claude: System Prompts— Hacker News

    AI 导读 · 提示工程新范式，系统提示词成AI行为核心控制手段。

  3. 15.
    Optima tackles AI benchmarking's biggest flaw by letting users test models against their own data— The Decoder

    AI 导读 · 自建数据测试模型，打破通用基准局限，让AI评估更贴近真实业务场景。

  4. 16.
    Top mathematicians say LLMs are strong calculators but poor creative thinkers— The Decoder

    AI 导读 · 大模型擅长组合旧方法，却难有真正数学直觉，揭示AI创新边界。

  5. 17.
    ChatGPT’s Computer History tracks your clicks and keystrokes— The Verge

    AI 导读 · 隐私与智能的边界再被突破，AI主动学习用户操作习惯，自动化潜力与数据安全堪忧。

  6. 18.
    Rogue AI aren’t science fiction anymore— The Verge

    AI 导读 · AI安全不再是科幻议题，监管与伦理成现实焦点。

论文研究4 条

  1. 19.
    One in five US workers now delegates tasks to AI instead of colleagues, survey finds— The Decoder

    A representative survey by Epoch AI found that 20 percent of employed Americans hand off at least one task to AI that a human used to do. Generally, they accept AI output with litt…

  2. 20.
    Meet Needle 2: An Open 45M-Parameter Tool-Calling Model That Ships as a 14MB Binary and Runs a Full Session in 28MB of RAM— MarkTechPost

    Cactus Compute released Needle 2, an open 45M-parameter model for tool calling, device use, and structured extraction. The full model is a single 14MB binary that runs a session in…

  3. 21.
    Liquid AI Releases LFM2.5-VL-3B: A 3B Vision-Language Model That Reads Screens, Grounds Objects, and Calls Tools On-Device— MarkTechPost

    Liquid AI released LFM2.5-VL-3B, a 3.1B-parameter vision-language model built for on-device deployment. It averages 80.7 on ScreenSpot-v2 and lifts RefCOCO grounding from 57.1 to 8…

  4. 22.
    AI宣布森多夫猜想告破！陶哲轩发现它隐藏的更强结果— 机器之心

    AI在基础科学研究的新进展。

技巧与观点6 条

  1. 23.
    Fine-Tuning Tool-Calling LLMs: A Complete Guide Using XYZ-Aquila-SFT and Qwen3— MarkTechPost

    Implement an end-to-end fine-tuning pipeline for tool-calling language models. This tutorial covers parsing trajectories, structured tool-call extraction, Qwen-compatible ChatML re…

  2. 24.
    Create a Reasoning-Focused LLM: A Practical Guide to Streaming, Curating, and Fine-Tuning the SupraLabs Reasoning Corpus— MarkTechPost

    This tutorial provides a complete workflow for building a compact, reasoning-focused language model. By streaming the SupraLabs reasoning corpus from Hugging Face, we apply quality…

  3. 25.
    AllenAI Open Instruct Tulu 3 Post-Training with SFT, DPO, RLVR, GRPO, and Verifier-Based Evaluation— MarkTechPost

    Build a custom LLM post-training pipeline using AllenAI’s Open Instruct framework. This comprehensive guide walks through Supervised Fine-Tuning (SFT), Direct Preference Optimizati…

  4. 26.
    Quoting Dario Amodei— Simon Willison

    AI 导读 · 公众对AI的负面看法主因不在警示，而在实际影响，值得深思。

  5. 27.
    CORS Chat— Simon Willison

    Tool: CORS Chat I built this today ( with GPT-5.6-Sol xhigh ) to help test Qwen 3.8 27B running in LM Studio on both my M5 MacBook Pro and an NVIDIA DGX Spark. It provides a web UI…

  6. 28.
    李飞飞最新访谈：AI咋能代替人呢？— 量子位

    AI 导读 · AI大佬亲述技术定位，破除替代焦虑，指明人机协作新方向。

快讯

日报生成时间：2026-09-06
← 2026-08-15最新日报2026-08-17 →