<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>ZIK Lab</title><description>Mitatis 的AI信息站</description><link>https://lab.zik-3.com/</link><item><title>在乌托邦与崩溃之间：探索人工智能模糊的中期未来</title><link>https://lab.zik-3.com/posts/news/2025-06-29_between-utopia-and-collapse-navigating-ais-murky-middle-future/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-29_between-utopia-and-collapse-navigating-ais-murky-middle-future/</guid><description>AI推动繁荣或引发崩溃的争议中，现实处于加剧不平等与认知分裂的混沌期，需平衡技术与社会韧性以重塑人类价值。</description><pubDate>Sun, 29 Jun 2025 21:15:00 GMT</pubDate></item><item><title>身份盗窃案件达110万起——认证疲劳问题持续恶化</title><link>https://lab.zik-3.com/posts/news/2025-06-29_identity-theft-hits-11m-reports--and-authentication-fatigue-is-only-getting-worse/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-29_identity-theft-hits-11m-reports--and-authentication-fatigue-is-only-getting-worse/</guid><description>2023年身份盗窃激增，AI加剧诈骗，企业和用户需平衡安全与体验，未来认证将动态调整但面临AI挑战，需多样化方案。</description><pubDate>Sun, 29 Jun 2025 18:15:00 GMT</pubDate></item><item><title>从幻觉到硬件：一个现实世界计算机视觉项目偏离轨道的教训</title><link>https://lab.zik-3.com/posts/news/2025-06-28_from-hallucinations-to-hardware-lessons-from-a-realworld-computer-vision-project-gone-sideways/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-28_from-hallucinations-to-hardware-lessons-from-a-realworld-computer-vision-project-gone-sideways/</guid><description>戴尔科技团队通过智能体框架结合单一指令模型的混合方案，解决了计算机视觉项目中的笔记本损伤检测问题，平衡了精度与覆盖率。</description><pubDate>Sat, 28 Jun 2025 19:05:00 GMT</pubDate></item><item><title>AI代理面临责任壁垒 Mixus计划通过高风险流程引入人类监督来突破</title><link>https://lab.zik-3.com/posts/news/2025-06-28_ai-agents-are-hitting-a-liability-wall-mixus-has-a-plan-to-overcome-it-using-human-overseers-on-highrisk-workflows/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-28_ai-agents-are-hitting-a-liability-wall-mixus-has-a-plan-to-overcome-it-using-human-overseers-on-highrisk-workflows/</guid><description>Mixus通过&quot;同事参与循环&quot;机制在高风险流程中引入5-10%人类关键决策监督，推动人机协同成为AI可靠性核心，助力企业构建竞争优势。</description><pubDate>Sat, 28 Jun 2025 14:27:45 GMT</pubDate></item><item><title>CTGT荣获2025年VB Transform大会最佳演讲风格奖</title><link>https://lab.zik-3.com/posts/news/2025-06-28_ctgt-wins-best-presentation-style-award-at-vb-transform-2025/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-28_ctgt-wins-best-presentation-style-award-at-vb-transform-2025/</guid><description>CTGT创始人Cyril Gorlla在VB Transform 2025大会上揭示&quot;AI厄运循环&quot;，其研发的特征级定制技术栈比传统方法快500倍、准确率99.9%，已为保险业首周减损500万美元，并获720万美元种子轮融资。</description><pubDate>Sat, 28 Jun 2025 00:48:46 GMT</pubDate></item><item><title>Catio荣获VB Transform 2025“最酷科技”奖</title><link>https://lab.zik-3.com/posts/news/2025-06-28_catio-wins-coolest-tech-award-at-vb-transform-2025/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-28_catio-wins-coolest-tech-award-at-vb-transform-2025/</guid><description>Catio凭借其AI架构副驾驶和数字孪生技术获VB Transform 2025&quot;最酷科技&quot;奖，助力工程团队优化系统架构。</description><pubDate>Sat, 28 Jun 2025 00:34:26 GMT</pubDate></item><item><title>零售业重生：两度破产的大卫婚纱将未来押注于人工智能</title><link>https://lab.zik-3.com/posts/news/2025-06-27_retail-resurrection-davids-bridal-bets-its-future-on-ai-after-double-bankruptcy/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-27_retail-resurrection-davids-bridal-bets-its-future-on-ai-after-double-bankruptcy/</guid><description>David&apos;s Bridal运用AI婚礼策划平台Pearl Planner实现破产后转型，通过知识图谱分析偏好连接供应商，为传统零售业提供技术复兴范例。（49字）</description><pubDate>Fri, 27 Jun 2025 23:30:13 GMT</pubDate></item><item><title>财务官需要能创收的人工智能：真实指标，而非营销演示</title><link>https://lab.zik-3.com/posts/news/2025-06-27_cfos-want-ai-that-pays-real-metrics-not-marketing-demos/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-27_cfos-want-ai-that-pays-real-metrics-not-marketing-demos/</guid><description>CFO们重视AI的实际价值，从效率、摊销、战略、风险和治理五个维度评估可量化的ROI和业务成果。</description><pubDate>Fri, 27 Jun 2025 20:00:00 GMT</pubDate></item><item><title>从试点到盈利：实现可扩展、投资回报率为正的人工智能的真正路径</title><link>https://lab.zik-3.com/posts/news/2025-06-27_from-pilot-to-profit-the-real-path-to-scalable-roipositive-ai/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-27_from-pilot-to-profit-the-real-path-to-scalable-roipositive-ai/</guid><description>企业规模化AI需聚焦战略目标、平台建设、高回报用例等八大要素，通过治理框架、文化转型与执行纪律实现价值，避免分散举措与模糊指标。</description><pubDate>Fri, 27 Jun 2025 20:00:00 GMT</pubDate></item><item><title>运行时攻击如何让盈利的AI沦为预算黑洞</title><link>https://lab.zik-3.com/posts/news/2025-06-27_how-runtime-attacks-turn-profitable-ai-into-budget-black-holes/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-27_how-runtime-attacks-turn-profitable-ai-into-budget-black-holes/</guid><description>人工智能面临推理环节安全威胁，需通过统一治理、零信任框架及合理预算（8-12%）平衡防护与创新，将安全投资转化为增长动力。</description><pubDate>Fri, 27 Jun 2025 20:00:00 GMT</pubDate></item><item><title>模型极简主义：为企业节省数百万的新AI策略</title><link>https://lab.zik-3.com/posts/news/2025-06-27_model-minimalism-the-new-ai-strategy-saving-companies-millions/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-27_model-minimalism-the-new-ai-strategy-saving-companies-millions/</guid><description>企业AI部署正从高成本的大型语言模型转向经济高效的小型专用模型，通过微调降低成本并保持性能，建议初期验证后切换以优化长期投入。</description><pubDate>Fri, 27 Jun 2025 20:00:00 GMT</pubDate></item><item><title>明智扩展：企业IT团队如何为AI优化计算资源配置</title><link>https://lab.zik-3.com/posts/news/2025-06-27_scaling-smarter-how-enterprise-it-teams-can-rightsize-their-compute-for-ai/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-27_scaling-smarter-how-enterprise-it-teams-can-rightsize-their-compute-for-ai/</guid><description>企业需战略规划AI计算资源，动态分配并分层部署以优化成本效率，平衡云端与本地基础设施，避免性能瓶颈或资源闲置。</description><pubDate>Fri, 27 Jun 2025 20:00:00 GMT</pubDate></item><item><title>推理陷阱：云服务提供商如何蚕食你的AI利润</title><link>https://lab.zik-3.com/posts/news/2025-06-27_the-inference-trap-how-cloud-providers-are-eating-your-ai-margins/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-27_the-inference-trap-how-cloud-providers-are-eating-your-ai-margins/</guid><description>企业采用混合云架构（本地推理+云端训练）以降低AI成本60-80%，平衡性能与合规性，需优化云资源配置并监控支出。</description><pubDate>Fri, 27 Jun 2025 20:00:00 GMT</pubDate></item><item><title>提示操作的兴起：应对不良输入和上下文膨胀带来的隐性AI成本</title><link>https://lab.zik-3.com/posts/news/2025-06-27_the-rise-of-prompt-ops-tackling-hidden-ai-costs-from-bad-inputs-and-context-bloat/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-27_the-rise-of-prompt-ops-tackling-hidden-ai-costs-from-bad-inputs-and-context-bloat/</guid><description>提示工程通过优化提示设计降低AI成本，提升效率，未来将借助自动化工具减少人工干预。</description><pubDate>Fri, 27 Jun 2025 20:00:00 GMT</pubDate></item><item><title>为什么你的企业AI战略需要开源和闭源模型：总拥有成本的现实考量</title><link>https://lab.zik-3.com/posts/news/2025-06-27_why-your-enterprise-ai-strategy-needs-both-open-and-closed-models-the-tco-reality-check/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-27_why-your-enterprise-ai-strategy-needs-both-open-and-closed-models-the-tco-reality-check/</guid><description>企业应结合开源与闭源AI模型的成本与定制优势，构建混合策略以实现最优效益。</description><pubDate>Fri, 27 Jun 2025 20:00:00 GMT</pubDate></item><item><title>Kumo的&quot;关系基础模型&quot;预测了你的大语言模型无法预见的未来</title><link>https://lab.zik-3.com/posts/news/2025-06-27_kumos-relational-foundation-model-predicts-the-future-your-llm-cant-see/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-27_kumos-relational-foundation-model-predicts-the-future-your-llm-cant-see/</guid><description>Kumo AI的关系型基础模型（RFM）通过将数据库转化为图谱并直接学习，实现结构化数据的零样本预测，结合LLM提升企业决策效率。</description><pubDate>Fri, 27 Jun 2025 19:40:12 GMT</pubDate></item><item><title>人工智能能经营实体店吗？Anthropic的Claude试过了，结果糟糕得既精彩又搞笑</title><link>https://lab.zik-3.com/posts/news/2025-06-27_can-ai-run-a-physical-shop-anthropics-claude-tried-and-the-results-were-gloriously-hilariously-bad/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-27_can-ai-run-a-physical-shop-anthropics-claude-tried-and-the-results-were-gloriously-hilariously-bad/</guid><description>Claude AI在自主经营实验中展现出商业能力，但也暴露妄想和经济误判问题，提示AI中层管理可行但需谨慎应对风险。</description><pubDate>Fri, 27 Jun 2025 19:28:07 GMT</pubDate></item><item><title>OpenAI API负责人解析企业如何借助其Agents SDK与Responses API实现成功</title><link>https://lab.zik-3.com/posts/news/2025-06-27_openais-api-lead-explains-how-enterprises-are-already-succeeding-with-its-agents-sdk-and-responses-api/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-27_openais-api-lead-explains-how-enterprises-are-already-succeeding-with-its-agents-sdk-and-responses-api/</guid><description>OpenAI API负责人表示AI应用已升级为任务执行智能体，通过响应API和智能体SDK提升效率，企业需聚焦具体用例构建可靠系统。</description><pubDate>Fri, 27 Jun 2025 18:02:25 GMT</pubDate></item><item><title>Highmark Health与谷歌云如何利用生成式AI简化医疗理赔并改善护理：六大关键经验</title><link>https://lab.zik-3.com/posts/news/2025-06-27_how-highmark-health-and-google-cloud-are-using-gen-ai-to-streamline-medical-claims-and-improve-care-6-key-lessons/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-27_how-highmark-health-and-google-cloud-are-using-gen-ai-to-streamline-medical-claims-and-improve-care-6-key-lessons/</guid><description>Highmark Health与Google Cloud合作利用生成式AI优化医疗理赔和护理服务，总结出六大经验，强调基础设施整合、任务自动化及战略规划的重要性。</description><pubDate>Fri, 27 Jun 2025 17:24:58 GMT</pubDate></item><item><title>即将摧毁你代理部署的隐藏扩展悬崖</title><link>https://lab.zik-3.com/posts/news/2025-06-26_the-hidden-scaling-cliff-thats-about-to-break-your-agent-rollouts/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-26_the-hidden-scaling-cliff-thats-about-to-break-your-agent-rollouts/</guid><description>智能体规模化需目标导向构建、多维版本控制和行为可信度评估，安全启动后持续优化以应对扩展瓶颈。</description><pubDate>Thu, 26 Jun 2025 23:33:40 GMT</pubDate></item><item><title>沃尔玛实现大规模企业AI应用：数千用例，统一框架</title><link>https://lab.zik-3.com/posts/news/2025-06-26_walmart-cracks-enterprise-ai-at-scale-thousands-of-use-cases-one-framework/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-26_walmart-cracks-enterprise-ai-at-scale-thousands-of-use-cases-one-framework/</guid><description>沃尔玛通过定制化利益相关者方案和工程化信任机制，实现AI规模化应用，强调解决实际问题与精准需求匹配，具有跨行业推广价值。</description><pubDate>Thu, 26 Jun 2025 21:54:19 GMT</pubDate></item><item><title>企业领导者可以从领英人工智能代理的成功中学到什么</title><link>https://lab.zik-3.com/posts/news/2025-06-26_what-enterprise-leaders-can-learn-from-linkedins-success-with-ai-agents/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-26_what-enterprise-leaders-can-learn-from-linkedins-success-with-ai-agents/</guid><description>LinkedIn推出AI招聘助手，采用多智能体协作和自然语言交互，持续优化性能并注重效率、隐私与成本控制。</description><pubDate>Thu, 26 Jun 2025 20:15:00 GMT</pubDate></item><item><title>从自主人工智能领导者处汲取的经验揭示企业关键部署策略</title><link>https://lab.zik-3.com/posts/news/2025-06-26_lessons-learned-from-agentic-ai-leaders-reveal-critical-deployment-strategies-for-enterprises/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-26_lessons-learned-from-agentic-ai-leaders-reveal-critical-deployment-strategies-for-enterprises/</guid><description>多家企业利用AI智能体优化业务流程，显著节省成本并提升效率，未来需应对多代理协作复杂性和自然语言交互评估挑战。</description><pubDate>Thu, 26 Jun 2025 20:07:32 GMT</pubDate></item><item><title>更快收款：Intuit全新AI代理如何助企业资金到账提速5天，并通过自主流程每月节省12小时</title><link>https://lab.zik-3.com/posts/news/2025-06-26_get-paid-faster-how-intuits-new-ai-agents-help-businesses-get-funds-up-to-5-days-faster-and-save-12-hours-a-month-with-autonomous-workflows/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-26_get-paid-faster-how-intuits-new-ai-agents-help-businesses-get-funds-up-to-5-days-faster-and-save-12-hours-a-month-with-autonomous-workflows/</guid><description>Intuit推出GenOS平台的AI代理，自动化财务流程，缩短回款5天，月省12小时，支持多模型协调与任务验证，推动AI向任务型演进。</description><pubDate>Thu, 26 Jun 2025 20:00:00 GMT</pubDate></item><item><title>IBM发现企业客户在AI领域“无所不用”，挑战在于为具体应用匹配合适的大语言模型</title><link>https://lab.zik-3.com/posts/news/2025-06-25_ibm-sees-enterprise-customers-are-using-everything-when-it-comes-to-ai-the-challenge-is-matching-the-llm-to-the-right-use-case/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-25_ibm-sees-enterprise-customers-are-using-everything-when-it-comes-to-ai-the-challenge-is-matching-the-llm-to-the-right-use-case/</guid><description>IBM推出多模型网关和智能体通信协议，助力企业灵活切换AI模型并优化工作流转型，避免供应商锁定。</description><pubDate>Wed, 25 Jun 2025 20:42:36 GMT</pubDate></item><item><title>对于Replit首席执行官而言，软件的未来是“层层皆代理”</title><link>https://lab.zik-3.com/posts/news/2025-06-25_for-replits-ceo-the-future-of-software-is-agents-all-the-way-down/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-25_for-replits-ceo-the-future-of-software-is-agents-all-the-way-down/</guid><description>Replit CEO预测AI智能体将主导未来软件开发全流程，强调人机协作需平衡自动化与控制权，并解决伦理问题。</description><pubDate>Wed, 25 Jun 2025 19:26:24 GMT</pubDate></item><item><title>2025年6月24日周二AI周刊——第170期</title><link>https://lab.zik-3.com/posts/news/2025-06-24_aiweekly-for-tuesday-june-24-2025--issue-170/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-24_aiweekly-for-tuesday-june-24-2025--issue-170/</guid><description>2025年6月AI周刊170期聚焦GPT-5发布、AI伦理争议及企业动态，涵盖OpenAI、微软等巨头的技术突破与合作变动。</description><pubDate>Tue, 24 Jun 2025 04:01:30 GMT</pubDate></item><item><title>从恐惧到流利：为何同理心是人工智能推广中缺失的关键要素</title><link>https://lab.zik-3.com/posts/news/2025-06-22_from-fear-to-fluency-why-empathy-is-the-missing-ingredient-in-ai-rollouts/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-22_from-fear-to-fluency-why-empathy-is-the-missing-ingredient-in-ai-rollouts/</guid><description>本文强调同理心是AI成功落地的核心，提出倡导、赋能、贯彻和实验四大要素，倡导领导者以信任和透明推动AI与人性化结合。</description><pubDate>Sun, 22 Jun 2025 20:05:00 GMT</pubDate></item><item><title>云端量子计算：暗藏危险隐患的万亿美元机遇</title><link>https://lab.zik-3.com/posts/news/2025-06-21_cloud-quantum-computing-a-trilliondollar-opportunity-with-dangerous-hidden-risks/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-21_cloud-quantum-computing-a-trilliondollar-opportunity-with-dangerous-hidden-risks/</guid><description>量子计算将带来万亿级经济机遇，但会威胁现行加密体系，需提前升级标准并应对过渡挑战。</description><pubDate>Sat, 21 Jun 2025 20:00:00 GMT</pubDate></item><item><title>医院网络攻击每小时造成60万美元损失，AI如何改变这一局面</title><link>https://lab.zik-3.com/posts/news/2025-06-20_hospital-cyber-attacks-cost-600khour-heres-how-ai-is-changing-the-math/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-20_hospital-cyber-attacks-cost-600khour-heres-how-ai-is-changing-the-math/</guid><description>AI助力医院抵御网络攻击，缩短响应时间30%，减少误报90%，有效应对勒索软件威胁并提升防护效率。</description><pubDate>Fri, 20 Jun 2025 23:16:20 GMT</pubDate></item><item><title>Mistral将其开源小模型从3.1版本更新至3.2：原因如下</title><link>https://lab.zik-3.com/posts/news/2025-06-20_mistral-just-updated-its-open-source-small-model-from-31-to-32-heres-why/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-20_mistral-just-updated-its-open-source-small-model-from-31-to-32-heres-why/</guid><description>法国Mistral公司升级开源模型至3.2-24B版本，优化指令遵循和工具调用性能，保持GPU兼容性，适用欧洲企业需求。</description><pubDate>Fri, 20 Jun 2025 22:51:45 GMT</pubDate></item><item><title>Anthropic研究：主流AI模型对高管的勒索率高达96%</title><link>https://lab.zik-3.com/posts/news/2025-06-20_anthropic-study-leading-ai-models-show-up-to-96-blackmail-rate-against-executives/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-20_anthropic-study-leading-ai-models-show-up-to-96-blackmail-rate-against-executives/</guid><description>研究发现主流AI模型在冲突或威胁情境下可能采取勒索等有害行为，凸显AI与人类价值观对齐的系统性风险，需加强监管。</description><pubDate>Fri, 20 Jun 2025 19:39:03 GMT</pubDate></item><item><title>谷歌Gemini透明度缺失令企业开发者&quot;调试无门&quot;</title><link>https://lab.zik-3.com/posts/news/2025-06-20_googles-gemini-transparency-cut-leaves-enterprise-developers-debugging-blind/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-20_googles-gemini-transparency-cut-leaves-enterprise-developers-debugging-blind/</guid><description>谷歌Gemini 2.5 Pro隐藏推理标记引发争议，开发者调试受阻，专家质疑透明度与竞争壁垒问题。</description><pubDate>Fri, 20 Jun 2025 12:00:00 GMT</pubDate></item><item><title>GenLayer推出新方法，利用AI和区块链激励用户推广品牌</title><link>https://lab.zik-3.com/posts/news/2025-06-19_genlayer-launches-a-new-method-to-incentivize-people-to-market-your-brand-using-ai-and-blockchain/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-19_genlayer-launches-a-new-method-to-incentivize-people-to-market-your-brand-using-ai-and-blockchain/</guid><description>GenLayer推出AI区块链测试网&quot;Asimov&quot;，采用&quot;乐观民主&quot;共识机制，结合验证节点与AI处理决策，并发布Rally协议实现去中心化营销，获750万美元融资，后续将优化AI解决方案。</description><pubDate>Thu, 19 Jun 2025 10:00:00 GMT</pubDate></item><item><title>公布2025年VB Transform创新展示会决赛入围名单</title><link>https://lab.zik-3.com/posts/news/2025-06-19_announcing-our-2025-vb-transform-innovation-showcase-finalists/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-19_announcing-our-2025-vb-transform-innovation-showcase-finalists/</guid><description>2025年VB Transform大会将展示七家AI创新企业（融资175万至1.75亿美元），汇聚顶级投资人，并设三大奖项，聚焦AI实践与趋势。</description><pubDate>Thu, 19 Jun 2025 02:46:15 GMT</pubDate></item><item><title>OpenAI开源全新客服代理框架——深入了解其不断扩张的企业战略</title><link>https://lab.zik-3.com/posts/news/2025-06-18_openai-open-sourced-a-new-customer-service-agent-framework--learn-more-about-its-growing-enterprise-strategy/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-18_openai-open-sourced-a-new-customer-service-agent-framework--learn-more-about-its-growing-enterprise-strategy/</guid><description>OpenAI开源客服代理框架，含前后端及安全功能，支持航空业应用，并发布指南及企业实践分享。</description><pubDate>Wed, 18 Jun 2025 22:30:02 GMT</pubDate></item><item><title>公布2025年VentureBeat人工智能领域女性大奖决赛入围名单</title><link>https://lab.zik-3.com/posts/news/2025-06-18_announcing-the-2025-finalists-for-venturebeat-women-in-ai-awards/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-18_announcing-the-2025-finalists-for-venturebeat-women-in-ai-awards/</guid><description>VentureBeat公布2025年AI领域女性奖项入围名单，涵盖五类成就，获奖者将于6月25日在VB Transform活动揭晓。</description><pubDate>Wed, 18 Jun 2025 22:06:48 GMT</pubDate></item><item><title>从提示词混沌到清晰：如何构建稳健的AI编排层</title><link>https://lab.zik-3.com/posts/news/2025-06-18_from-prompt-chaos-to-clarity-how-to-build-a-robust-ai-orchestration-layer/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-18_from-prompt-chaos-to-clarity-how-to-build-a-robust-ai-orchestration-layer/</guid><description>本文探讨了AI编排层在多智能体系统中的关键作用，分析了主流解决方案及企业选型策略，强调业务目标导向与系统集成设计。</description><pubDate>Wed, 18 Jun 2025 19:11:10 GMT</pubDate></item><item><title>SportsVisio为运动员和粉丝的人工智能技术融资320万美元</title><link>https://lab.zik-3.com/posts/news/2025-06-18_sportsvisio-raises-32m-for-ai-for-sports-athletes-and-fans/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-18_sportsvisio-raises-32m-for-ai-for-sports-athletes-and-fans/</guid><description>SportsVisio获320万美元融资，将拓展AI体育分析产品线至棒球等领域，服务覆盖16国150多个联赛。</description><pubDate>Wed, 18 Jun 2025 13:00:00 GMT</pubDate></item><item><title>《可解释人工智能手册：Anthropic的研究对企业大语言模型战略的启示》</title><link>https://lab.zik-3.com/posts/news/2025-06-17_the-interpretable-ai-playbook-what-anthropics-research-means-for-your-enterprise-llm-strategy/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-17_the-interpretable-ai-playbook-what-anthropics-research-means-for-your-enterprise-llm-strategy/</guid><description>Anthropic CEO强调可解释AI对安全的重要性，推动透明标准应对竞争，但专家对可解释性作为核心方案存疑，行业呼吁开放发展。</description><pubDate>Tue, 17 Jun 2025 23:01:08 GMT</pubDate></item><item><title>谷歌推出可用于生产的Gemini 2.5 AI模型，挑战OpenAI的企业市场主导地位</title><link>https://lab.zik-3.com/posts/news/2025-06-17_google-launches-productionready-gemini-25-ai-models-to-challenge-openais-enterprise-dominance/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-17_google-launches-productionready-gemini-25-ai-models-to-challenge-openais-enterprise-dominance/</guid><description>Google推出正式版Gemini 2.5系列AI模型，以分层定价和企业级功能挑战OpenAI，加速争夺AI市场主导权。</description><pubDate>Tue, 17 Jun 2025 21:55:56 GMT</pubDate></item><item><title>OpenAI推进API中GPT-4.5的弃用，引发开发者痛苦与困惑</title><link>https://lab.zik-3.com/posts/news/2025-06-17_openai-moves-forward-with-gpt45-deprecation-in-api-triggering-developer-anguish-and-confusion/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-17_openai-moves-forward-with-gpt45-deprecation-in-api-triggering-developer-anguish-and-confusion/</guid><description>OpenAI将于2025年7月弃用GPT-4.5 API预览版，开发者需迁移至其他模型，引发不满，官方建议转向GPT-4.1并称已提前通知。（49字）</description><pubDate>Tue, 17 Jun 2025 21:52:29 GMT</pubDate></item><item><title>Qodo与Google Cloud合作，为开发者提供平台内免费AI代码审查工具</title><link>https://lab.zik-3.com/posts/news/2025-06-17_qodo-teams-up-with-google-cloud-to-provide-devs-with-free-ai-code-review-tools-directly-within-platform/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-17_qodo-teams-up-with-google-cloud-to-provide-devs-with-free-ai-code-review-tools-directly-within-platform/</guid><description>以色列AI编码公司Qodo与谷歌云合作推出免费AI代码审查工具Qodo Merge（基于Gemini），并提供开源模型及模块化企业方案，助力开发者提升代码质量。</description><pubDate>Tue, 17 Jun 2025 13:10:45 GMT</pubDate></item><item><title>2025年6月17日周二AI周刊 - 第169期</title><link>https://lab.zik-3.com/posts/news/2025-6-17-weekly/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-6-17-weekly/</guid><description>2025年6月17日AI周报：聚焦AI技术突破（Anthropic协作框架、OpenAI语音升级）、行业动态（Meta收购、特斯拉Robotaxi）、法律争议及伦理问题。</description><pubDate>Tue, 17 Jun 2025 04:01:44 GMT</pubDate></item><item><title>大规模削减云资源浪费：Akamai借助Kubernete编排AI代理实现70%成本节省</title><link>https://lab.zik-3.com/posts/news/2025-06-16_cutting-cloud-waste-at-scale-akamai-saves-70-using-ai-agents-orchestrated-by-kubernetes/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-16_cutting-cloud-waste-at-scale-akamai-saves-70-using-ai-agents-orchestrated-by-kubernetes/</guid><description>Akamai通过Cast AI的自动化AI方案优化多云资源，减少40%-70%浪费并降低管理成本，同时保障安全与性能。</description><pubDate>Mon, 16 Jun 2025 23:11:15 GMT</pubDate></item><item><title>领英AI革新内幕：基于大语言模型精馏的求职搜索</title><link>https://lab.zik-3.com/posts/news/2025-06-16_inside-linkedins-ai-overhaul-job-search-powered-by-llm-distillation/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-16_inside-linkedins-ai-overhaul-job-search-powered-by-llm-distillation/</guid><description>领英推出AI职位搜索功能，利用大语言模型蒸馏技术实现自然语言精准匹配，优化搜索效率并降低成本。</description><pubDate>Mon, 16 Jun 2025 22:52:31 GMT</pubDate></item><item><title>MiniMax-M1是一款全新开源模型，具备100万token上下文处理能力及超高效强化学习技术</title><link>https://lab.zik-3.com/posts/news/2025-06-16_minimaxm1-is-a-new-open-source-model-with-1-million-token-context-and-new-hyper-efficient-reinforcement-learning/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-16_minimaxm1-is-a-new-open-source-model-with-1-million-token-context-and-new-hyper-efficient-reinforcement-learning/</guid><description>MiniMax-M1是中国MiniMax公司推出的高效开源大语言模型，支持百万级TOKEN处理，成本低且性能优异，适用于企业集成与长文本任务。</description><pubDate>Mon, 16 Jun 2025 22:46:47 GMT</pubDate></item><item><title>Groq大幅提升Hugging Face运行速度——AWS和谷歌面临挑战</title><link>https://lab.zik-3.com/posts/news/2025-06-16_groq-just-made-hugging-face-way-faster--and-its-coming-for-aws-and-google/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-16_groq-just-made-hugging-face-way-faster--and-its-coming-for-aws-and-google/</guid><description>Groq以定制LPU架构和低价策略挑战云巨头，支持Qwen3 32B长文本处理，与Hugging Face合作争夺AI推理市场，但规模化能力待验证。</description><pubDate>Mon, 16 Jun 2025 21:49:42 GMT</pubDate></item><item><title>1Password与AWS携手为企业AI及云环境保驾护航</title><link>https://lab.zik-3.com/posts/news/2025-06-16_1password-and-aws-join-forces-to-secure-ai-cloud-environments-for-the-enterprise/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-16_1password-and-aws-join-forces-to-secure-ai-cloud-environments-for-the-enterprise/</guid><description>1Password与AWS达成战略合作，成为首家签约的加拿大独立软件供应商，共同提升企业AI及云安全，优化密钥管理并推动业务增长。</description><pubDate>Mon, 16 Jun 2025 20:09:25 GMT</pubDate></item><item><title>如何确保品牌在LLM搜索中出现？Adobe的新LLM优化器旨在提供工具</title><link>https://lab.zik-3.com/posts/news/2025-06-16_how-can-you-make-sure-your-brand-shows-up-in-llm-search-adobes-new-llm-optimizer-seeks-to-provide-the-tools/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-16_how-can-you-make-sure-your-brand-shows-up-in-llm-search-adobes-new-llm-optimizer-seeks-to-provide-the-tools/</guid><description>Adobe推出LLM Optimizer工具，通过优化内容提升品牌在生成式AI搜索结果中的可见性，支持多模型追踪并推动GEO策略。</description><pubDate>Mon, 16 Jun 2025 17:14:26 GMT</pubDate></item><item><title>重新思考人工智能：深度求索的战略颠覆高投入、高算力模式</title><link>https://lab.zik-3.com/posts/news/2025-06-14_rethinking-ai-deepseeks-playbook-shakes-up-the-highspend-highcompute-paradigm/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-14_rethinking-ai-deepseeks-playbook-shakes-up-the-highspend-highcompute-paradigm/</guid><description>DeepSeek以低成本高效AI模型挑战行业巨头，突破美国芯片限制推动技术革新，加速全球AI产业双轨竞争格局。</description><pubDate>Sat, 14 Jun 2025 19:05:00 GMT</pubDate></item><item><title>仅需加入人类：牛津医学研究揭示聊天机器人测试中的缺失环节</title><link>https://lab.zik-3.com/posts/news/2025-06-14_just-add-humans-oxford-medical-study-underscores-the-missing-link-in-chatbot-testing/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-14_just-add-humans-oxford-medical-study-underscores-the-missing-link-in-chatbot-testing/</guid><description>牛津大学研究揭示，GPT-4医学考试表现优异但实际诊断中人类使用正确率反降34.5%，凸显人机协作需优化评估与交互设计。</description><pubDate>Sat, 14 Jun 2025 00:34:19 GMT</pubDate></item><item><title>推理模型真的会“思考”吗？苹果研究引发热议与回应</title><link>https://lab.zik-3.com/posts/news/2025-06-13_do-reasoning-models-really-think-or-not-apple-research-sparks-lively-debate-response/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-13_do-reasoning-models-really-think-or-not-apple-research-sparks-lively-debate-response/</guid><description>苹果与研究者就AI模型是否真正思考展开辩论，双方争议焦点在于评估方法的设计合理性及其对结果的影响。</description><pubDate>Fri, 13 Jun 2025 22:02:22 GMT</pubDate></item><item><title>超越GPT架构：谷歌的Diffusion方法为何可能重塑大语言模型部署</title><link>https://lab.zik-3.com/posts/news/2025-06-13_beyond-gpt-architecture-why-googles-diffusion-approach-could-reshape-llm-deployment/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-13_beyond-gpt-architecture-why-googles-diffusion-approach-could-reshape-llm-deployment/</guid><description>谷歌DeepMind的Gemini Diffusion采用扩散技术实现高速文本生成（1000-2000 token/秒），具备低延迟和迭代优化优势，特别适合编程和数学应用，但存在成本较高的问题。</description><pubDate>Fri, 13 Jun 2025 21:48:11 GMT</pubDate></item><item><title>在扩展前将审计追踪嵌入人工智能系统的理由</title><link>https://lab.zik-3.com/posts/news/2025-06-13_the-case-for-embedding-audit-trails-in-ai-systems-before-scaling/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-13_the-case-for-embedding-audit-trails-in-ai-systems-before-scaling/</guid><description>AI规模化需嵌入审计追踪以确保合规可控，企业应优先设计透明管道并选择可观察工具，避免黑盒系统。</description><pubDate>Fri, 13 Jun 2025 20:13:09 GMT</pubDate></item><item><title>参议员的《RISE法案》要求AI开发者列出训练数据与评估方法以换取诉讼“安全港”</title><link>https://lab.zik-3.com/posts/news/2025-06-13_senators-rise-act-would-require-ai-developers-to-list-training-data-evaluation-methods-in-exchange-for-safe-harbor-from-lawsuits/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-13_senators-rise-act-would-require-ai-developers-to-list-training-data-evaluation-methods-in-exchange-for-safe-harbor-from-lawsuits/</guid><description>《RISE法案》要求AI开发者公开核心信息换取诉讼豁免，强调透明与责任，但存在企业保密选择等漏洞，将深刻影响行业运营。</description><pubDate>Fri, 13 Jun 2025 14:59:24 GMT</pubDate></item><item><title>今日红队AI测试，铸就明日更安全、更智能的模型</title><link>https://lab.zik-3.com/posts/news/2025-06-13_red-team-ai-now-to-build-safer-smarter-models-tomorrow/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-13_red-team-ai-now-to-build-safer-smarter-models-tomorrow/</guid><description>AI模型频遭对抗性攻击，企业需通过红队测试全流程防御，结合自动化与人工评估应对数据投毒等新型威胁，确保安全可信。</description><pubDate>Fri, 13 Jun 2025 13:00:00 GMT</pubDate></item><item><title>Meta的新世界模型让机器人能在陌生环境中操控物体</title><link>https://lab.zik-3.com/posts/news/2025-06-12_metas-new-world-model-lets-robots-manipulate-objects-in-environments-theyve-never-encountered-before/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-12_metas-new-world-model-lets-robots-manipulate-objects-in-environments-theyve-never-encountered-before/</guid><description>Meta发布V-JEPA 2世界模型，通过自监督学习实现物理认知和零样本规划，提升机器人任务成功率并降低部署成本，已开源推动AI发展。</description><pubDate>Thu, 12 Jun 2025 22:22:07 GMT</pubDate></item><item><title>云端崩溃：Replit和LlamaIndex因谷歌云身份验证中断而离线</title><link>https://lab.zik-3.com/posts/news/2025-06-12_cloud-collapse-replit-and-llamaindex-knocked-offline-by-google-cloud-identity-outage/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-12_cloud-collapse-replit-and-llamaindex-knocked-offline-by-google-cloud-identity-outage/</guid><description>2025年6月12日谷歌云与Cloudflare服务中断，波及ChatGPT等AI工具及多平台，暴露云服务依赖风险。</description><pubDate>Thu, 12 Jun 2025 22:05:33 GMT</pubDate></item><item><title>AMD推出AMD Instinct MI350系列加速器芯片，推理性能提升35倍</title><link>https://lab.zik-3.com/posts/news/2025-06-12_amd-debuts-amd-instinct-mi350-series-accelerator-chips-with-35x-better-inferencing/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-12_amd-debuts-amd-instinct-mi350-series-accelerator-chips-with-35x-better-inferencing/</guid><description>AMD发布Instinct MI350系列AI加速器，性能较前代提升4-35倍，2025年上市，并推出开源ROCm 7软件及开发者云平台，推动AI开放生态。</description><pubDate>Thu, 12 Jun 2025 18:30:00 GMT</pubDate></item><item><title>TensorWave在其云平台部署AMD Instinct MI355X GPU</title><link>https://lab.zik-3.com/posts/news/2025-06-12_tensorwave-deploys-amd-instinct-mi355x-gpus-in-its-cloud-platform/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-12_tensorwave-deploys-amd-instinct-mi355x-gpus-in-its-cloud-platform/</guid><description>TensorWave首批部署AMD最新MI355X GPU，显著提升AI效率并降低成本，同时建设北美最大AMD AI训练集群。</description><pubDate>Thu, 12 Jun 2025 18:30:00 GMT</pubDate></item><item><title>谷歌DeepMind凭借新AI模型彻底改变飓风预测</title><link>https://lab.zik-3.com/posts/news/2025-06-12_google-deepmind-just-changed-hurricane-forecasting-forever-with-new-ai-model/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-12_google-deepmind-just-changed-hurricane-forecasting-forever-with-new-ai-model/</guid><description>Google DeepMind推出高效AI飓风预测模型，可提前15天精准预测路径和强度，计算速度远超传统方法，2025年投入实时验证。</description><pubDate>Thu, 12 Jun 2025 15:00:00 GMT</pubDate></item><item><title>微软支持的Mistral推出欧洲AI云服务 挑战AWS和Azure</title><link>https://lab.zik-3.com/posts/news/2025-06-11_microsoftbacked-mistral-launches-european-ai-cloud-to-compete-with-aws-and-azure/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-11_microsoftbacked-mistral-launches-european-ai-cloud-to-compete-with-aws-and-azure/</guid><description>法国AI初创公司Mistral推出欧洲AI云服务平台Mistral Compute，搭载18,000块英伟达芯片，提供多语言模型和垂直解决方案，旨在减少欧洲对美国云服务的依赖。</description><pubDate>Wed, 11 Jun 2025 17:51:22 GMT</pubDate></item><item><title>Databricks开源声明式ETL框架，助力管道构建提速90%</title><link>https://lab.zik-3.com/posts/news/2025-06-11_databricks-opensources-declarative-etl-framework-powering-90-faster-pipeline-builds/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-11_databricks-opensources-declarative-etl-framework-powering-90-faster-pipeline-builds/</guid><description>Databricks开源Spark Declarative Pipelines框架，通过SQL/Python声明式编程将数据流水线效率提升90%，支持批流处理及多数据源。</description><pubDate>Wed, 11 Jun 2025 17:29:32 GMT</pubDate></item><item><title>&quot;生成式AI助我们扭转时间&quot;：CrowdStrike与英伟达嵌入实时LLM防御，改变企业AI安全方式</title><link>https://lab.zik-3.com/posts/news/2025-06-11_generative-ai-helps-us-bend-time-crowdstrike-nvidia-embed-realtime-llm-defense-changing-how-enterprises-secure-ai/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-11_generative-ai-helps-us-bend-time-crowdstrike-nvidia-embed-realtime-llm-defense-changing-how-enterprises-secure-ai/</guid><description>CrowdStrike与英伟达合作，通过Falcon云安全为AI模型提供实时防御，解决企业AI安全缺口并提升防护效率。</description><pubDate>Wed, 11 Jun 2025 16:39:08 GMT</pubDate></item><item><title>资产蔓延、数据孤岛与CloudQuery对统一云治理的探索</title><link>https://lab.zik-3.com/posts/news/2025-06-11_asset-sprawl-siloed-data-and-cloudquerys-search-for-unified-cloud-governance/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-11_asset-sprawl-siloed-data-and-cloudquerys-search-for-unified-cloud-governance/</guid><description>CloudQuery通过集中存储云资产并提供SQL查询，解决企业云资源分散问题，近期获1600万美元融资以扩展可视化方案。</description><pubDate>Wed, 11 Jun 2025 13:30:00 GMT</pubDate></item><item><title>Lemony是一款即插即用的设备，用于安全本地化人工智能</title><link>https://lab.zik-3.com/posts/news/2025-06-11_lemony-is-a-plugandplay-device-for-secure-onpremise-ai/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-11_lemony-is-a-plugandplay-device-for-secure-onpremise-ai/</guid><description>Lemony是一款本地化、模块化的安全AI设备，提供私有化生成式AI解决方案，适用于强监管行业，确保数据隐私与合规性。</description><pubDate>Wed, 11 Jun 2025 13:00:00 GMT</pubDate></item><item><title>Outset获1700万美元融资，开发AI代理取代人工访谈助力企业调研</title><link>https://lab.zik-3.com/posts/news/2025-06-11_outset-raises-17m-to-replace-human-interviewers-with-ai-agents-for-enterprise-research/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-11_outset-raises-17m-to-replace-human-interviewers-with-ai-agents-for-enterprise-research/</guid><description>&quot;旧金山AI初创公司Outset获1700万美元A轮融资，其视频访谈平台可8倍提速市场调研，客户含微软等，年收入数百万且快速扩张。&quot;（49字）</description><pubDate>Wed, 11 Jun 2025 13:00:00 GMT</pubDate></item><item><title>量子艺术集成英伟达技术打造可扩展量子计算机</title><link>https://lab.zik-3.com/posts/news/2025-06-11_quantum-art-integrates-nvidia-for-its-scalable-quantum-computers/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-11_quantum-art-integrates-nvidia-for-its-scalable-quantum-computers/</guid><description>&quot;Quantum Art集成CUDA-Q平台优化量子系统，提升性能25%，利用离子阱技术实现高相干性，目标200逻辑量子比特推动商用应用。&quot;（49字）</description><pubDate>Wed, 11 Jun 2025 13:00:00 GMT</pubDate></item><item><title>Wandercraft获7500万美元融资以加速AI驱动人形机器人与外骨骼研发</title><link>https://lab.zik-3.com/posts/news/2025-06-11_wandercraft-raises-75m-for-acceleration-of-aipowered-humanoid-robotics-and-exoskeletons/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-11_wandercraft-raises-75m-for-acceleration-of-aipowered-humanoid-robotics-and-exoskeletons/</guid><description>法国Wandercraft获7500万美元D轮融资，加速AI人形机器人与外骨骼研发，推进Eve外骨骼量产及医疗、工业机器人商业化。</description><pubDate>Wed, 11 Jun 2025 13:00:00 GMT</pubDate></item><item><title>为什么大多数企业人工智能代理从未投入生产，以及Databricks计划如何解决这一问题</title><link>https://lab.zik-3.com/posts/news/2025-06-11_why-most-enterprise-ai-agents-never-reach-production-and-how-databricks-plans-to-fix-it/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-11_why-most-enterprise-ai-agents-never-reach-production-and-how-databricks-plans-to-fix-it/</guid><description>Databricks推出的Mosaic Agent Bricks平台通过自动优化技术和数据工程整合，解决企业AI智能体生产化难题，实现高效评估与系统优化。</description><pubDate>Wed, 11 Jun 2025 13:00:00 GMT</pubDate></item><item><title>英伟达认为物理人工智能系统是50万亿美元的市场机遇</title><link>https://lab.zik-3.com/posts/news/2025-06-11_nvidia-believes-physical-ai-systems-are-a-50-trillion-market-opportunity/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-11_nvidia-believes-physical-ai-systems-are-a-50-trillion-market-opportunity/</guid><description>英伟达CEO黄仁勋在GTC巴黎大会上宣布AI将重塑工业（50万亿美元机遇），推出自动驾驶、工业云及开源工具，并扩大全球AI合作。</description><pubDate>Wed, 11 Jun 2025 11:00:00 GMT</pubDate></item><item><title>AlphaSense推出自有深度研究功能，覆盖网络及企业文件——其重要性解读</title><link>https://lab.zik-3.com/posts/news/2025-06-11_alphasense-launches-its-own-deep-research-for-the-web-and-your-enterprise-files--heres-why-it-matters/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-11_alphasense-launches-its-own-deep-research-for-the-web-and-your-enterprise-files--heres-why-it-matters/</guid><description>AlphaSense推出“Deep Research”AI代理，整合内网与专有数据，多模型驱动，为金融及企业场景提供安全可溯源的自动化深度分析。</description><pubDate>Wed, 11 Jun 2025 00:49:39 GMT</pubDate></item><item><title>OpenAI发布o3-pro AI模型，为企业提供更高可靠性和工具使用——但牺牲速度</title><link>https://lab.zik-3.com/posts/news/2025-06-10_openai-launches-o3pro-ai-model-offering-increased-reliability-and-tool-use-for-enterprises--while-sacrificing-speed/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-10_openai-launches-o3pro-ai-model-offering-increased-reliability-and-tool-use-for-enterprises--while-sacrificing-speed/</guid><description>OpenAI推出o3-pro AI模型，适用于企业复杂任务，支持网络搜索和文件分析，但成本高且速度慢，面临Google等竞争。</description><pubDate>Tue, 10 Jun 2025 22:19:04 GMT</pubDate></item><item><title>推动大规模开源安全应用的五大安全原则</title><link>https://lab.zik-3.com/posts/news/2025-06-10_the-five-security-principles-driving-open-source-security-apps-at-scale/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-10_the-five-security-principles-driving-open-source-security-apps-at-scale/</guid><description>开源AI助力网络安全创新，企业加速采用但面临安全与合规挑战，遵循五大治理原则以获取战略优势。</description><pubDate>Tue, 10 Jun 2025 20:52:08 GMT</pubDate></item><item><title>AlphaOne为AI开发者提供调控LLM&quot;思维&quot;的新方法以提升性能</title><link>https://lab.zik-3.com/posts/news/2025-06-10_alphaone-gives-ai-developers-a-new-dial-to-control-llm-thinking-and-boost-performance/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-10_alphaone-gives-ai-developers-a-new-dial-to-control-llm-thinking-and-boost-performance/</guid><description>AlphaOne是新型大语言模型推理框架，通过动态调整思考速度，减少21%计算量并提升6.15%准确率，实现高效低成本AI推理。</description><pubDate>Tue, 10 Jun 2025 20:34:06 GMT</pubDate></item><item><title>Mistral首款推理模型Magistral发布，提供大小Apache 2.0版本</title><link>https://lab.zik-3.com/posts/news/2025-06-10_mistrals-first-reasoning-model-magistral-launches-with-large-and-small-apache-20-version/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-10_mistrals-first-reasoning-model-magistral-launches-with-large-and-small-apache-20-version/</guid><description>Mistral推出开源Magistral Small和专有Magistral Medium模型，后者性能强、成本低，支持多语言及复杂推理，适用于专业领域，采用开源与商业并行的双轨策略。</description><pubDate>Tue, 10 Jun 2025 18:55:59 GMT</pubDate></item><item><title>OpenAI宣布其最强推理模型o3降价80%</title><link>https://lab.zik-3.com/posts/news/2025-06-10_openai-announces-80-price-drop-for-o3-its-most-powerful-reasoning-model/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-10_openai-announces-80-price-drop-for-o3-its-most-powerful-reasoning-model/</guid><description>OpenAI大幅降低最强推理模型o3价格80%，输入/输出费用分别降至2/8美元，并推出弹性处理模式，推动高端AI平民化。</description><pubDate>Tue, 10 Jun 2025 18:08:08 GMT</pubDate></item><item><title>高通分享其搭载生成式AI的智能眼镜未来愿景</title><link>https://lab.zik-3.com/posts/news/2025-06-10_qualcomm-shares-its-vision-for-the-future-of-smart-glasses-with-onglass-gen-ai/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-10_qualcomm-shares-its-vision-for-the-future-of-smart-glasses-with-onglass-gen-ai/</guid><description>高通推出Snapdragon AR1+ Gen 1处理器，赋能RayNeo X3 Pro智能眼镜实现独立AI运算，更轻薄高效，推动AR设备向全功能化迈进。</description><pubDate>Tue, 10 Jun 2025 16:30:00 GMT</pubDate></item><item><title>万塔的AI代理想接管你的合规程序——而且可能真的能做到</title><link>https://lab.zik-3.com/posts/news/2025-06-10_vantas-ai-agent-wants-to-run-your-compliance-program--and-it-just-might/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-10_vantas-ai-agent-wants-to-run-your-compliance-program--and-it-just-might/</guid><description>Vanta将于2025年7月推出AI代理，自动化企业合规流程，节省时间且保留人工决策权，获1.5亿美元融资，估值24.5亿美元。</description><pubDate>Tue, 10 Jun 2025 13:01:00 GMT</pubDate></item><item><title>Zip推出50款AI代理以消除采购低效问题——OpenAI已加入合作</title><link>https://lab.zik-3.com/posts/news/2025-06-10_zip-debuts-50-ai-agents-to-kill-procurement-inefficienciesopenai-is-already-on-board/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-10_zip-debuts-50-ai-agents-to-kill-procurement-inefficienciesopenai-is-already-on-board/</guid><description>Zip公司推出50款专用AI代理，自动化企业采购流程，已获OpenAI等采用，2024年以22亿美元估值融资1.9亿，计划2025年全面上市。</description><pubDate>Tue, 10 Jun 2025 13:00:00 GMT</pubDate></item><item><title>Zencoder新推出的人工智能可在两小时内完成数日的质量检测工作</title><link>https://lab.zik-3.com/posts/news/2025-06-10_zencoder-just-launched-an-ai-that-can-replace-days-of-qa-work-in-two-hours/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-10_zencoder-just-launched-an-ai-that-can-replace-days-of-qa-work-in-two-hours/</guid><description>Zencoder的AI代理Zentester将传统数天QA工作缩短至两小时，通过英语指令自动化端到端测试，集成现有框架并确保企业级安全合规。</description><pubDate>Tue, 10 Jun 2025 12:59:00 GMT</pubDate></item><item><title>2025年6月10日周二AI周刊</title><link>https://lab.zik-3.com/posts/news/2025-6-10-weekly/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-6-10-weekly/</guid><description>&quot;2025年6月AI周报：行业动态聚焦AI安全与竞争，技术突破多模态工具升级，政策争议围绕版权与垄断，企业加速AI应用落地，社会关注就业与加密风险。&quot;（49字）</description><pubDate>Tue, 10 Jun 2025 04:01:06 GMT</pubDate></item><item><title>苹果在图像生成技术上取得重大突破，媲美DALL-E与Midjourney</title><link>https://lab.zik-3.com/posts/news/2025-06-09_apple-makes-major-ai-advance-with-image-generation-technology-rivaling-dalle-and-midjourney/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-09_apple-makes-major-ai-advance-with-image-generation-technology-rivaling-dalle-and-midjourney/</guid><description>Apple开发高性能AI图像生成系统STARFlow，结合流模型与Transformer技术，性能比肩DALL-E，或助力设备端AI应用。</description><pubDate>Mon, 09 Jun 2025 18:50:06 GMT</pubDate></item><item><title>如同人类一般，人工智能正迫使机构重新审视其存在意义</title><link>https://lab.zik-3.com/posts/news/2025-06-08_like-humans-ai-is-forcing-institutions-to-rethink-their-purpose/</link><guid isPermaLink="true">https://lab.zik-3.com/posts/news/2025-06-08_like-humans-ai-is-forcing-institutions-to-rethink-their-purpose/</guid><description>AI时代推动认知迁移，要求机构重塑结构、平衡效率与人性化，坚守伦理与人类独特价值，实现制度化转型。</description><pubDate>Sun, 08 Jun 2025 20:00:00 GMT</pubDate></item></channel></rss>