{ "text": "大家好，这里是最佳拍档，我是大飞\n当我们每天使用AI生成文案、图片甚至视频的时候\n是否想过，这些看似神奇的技术背后\n正运行着一个由数据、算力和资本构成的庞大帝国\n今天\n我们将回顾《华尔街日报》科技记者郝珂灵（Karen Hao）在Novara Media的一场专访\n看看她是如何通过对硅谷AI行业的深度调查\n揭开那些推动技术进步的关键力量如何重塑世界\n以及这场技术革命背后隐藏的系统性风险\n这期视频我们将从概念的起源到资源的争夺\n从企业的蜕变到全球供应链的暗角\n深入了解这个正在成型的\"AI帝国\"，\n究竟是如何运作的\n要想理解当前AI产业的疯狂扩张\n必须先厘清\"人工智能\"这个充满模糊的概念\n1956年\n达特茅斯学院的助理教授约翰·麦卡锡（John McCarthy）\n为了给夏季研究项目筹集资金\n首次创造了\"人工智能\"这个术语\n这个诞生于学术营销的词汇\n从此成为了一个可以装入任何技术的\"手提箱概念\"。\n如今OpenAI、谷歌等巨头所谈论的AI\n本质上是机器学习的一个子类，即深度学习\n也就是通过神经网络对海量的数据进行模式计算\n实现文本生成、图像识别等功能\n但是公众很容易将它等同于科幻作品中的通用智能\n忽视了它作为一种特定技术工具的本质\n这种概念上的模糊\n导致了严重的认知偏差\n比如当英国政府宣称\n用AI为英国国家医疗服务体系节省了20%开支\"时\n并没有说明具体用的是哪种AI技术\n究竟是基于规则的医疗影像分析\n还是需要海量数据训练的大语言模型？\n两种技术在成本、效果和实施难度上的差异巨大\n但是\"人工智能\"这个笼统的表述掩盖了所有的细节\n更关键的是\n这种模糊性让公众难以评估技术的真实影响\n比如很少有人意识到\n训练一个大语言模型的碳排放量\n相当于数十万辆汽车的终身排放量\nAI技术的每一次突破\n都伴随着算力需求的指数级增长\n麦肯锡报告显示\n未来五年全球AI数据中心的能源消耗\n将每年新增相当于英国0.5到1.2倍的用电量\n而其中60%以上要依赖煤炭、天然气等化石燃料\n在得克萨斯州\n多家本应退役的燃煤电厂被重新启用\n唯一的目的就是为AI训练提供稳定的电力；\n埃隆·马斯克（Elon Musk）的XAI公司在田纳西州孟菲斯建设的Colossus超级计算机\n未经许可便安装了35台甲烷燃气轮机\n这些设备每天排放的有毒污染物超过了当地工业标准的200%，\n严重威胁到周边居民的健康\n水资源的争夺同样激烈\n由于数据中心需要纯净水冷却设备\n全球67%的新建AI数据中心\n选址在干旱或者半干旱地区\n乌拉圭首都蒙得维的亚正在经历百年一遇的干旱\n当地政府不得不将处理后的工业废水混入饮用水\n但是谷歌仍计划在此建设新的数据中心\n该项目预计每天将消耗8万吨淡水\n相当于16万居民的日常用水量\n这种\"技术发展优先\"的逻辑\n正在将环境成本转嫁到公共资源和弱势群体身上\n郝珂灵指出，当前美国AI领域的竞争\n本质上是一场关于\"通用人工智能（AGI）\"的集体狂欢\nOpenAI、Anthropic、谷歌、Meta等公司投入数百亿美元\n声称要创造\"重塑人类智能\"的技术\n但是颇具讽刺意味的是\n这些公司的商业案例普遍都是模糊不清的\n比如微软在2024年悄悄撤回了部分数据中心的投资\n承认\"短期回报不及预期\"，\n而OpenAI向投资者描绘的\"AGI垄断\"愿景\n更多是基于彼得·蒂尔（Peter Thiel）「竞争是给白痴准备的，创业必须要垄断」的理论\n创业必须要垄断\"的理论\n而非实际的技术路径\n与美国的概念驱动不同\n中国AI企业展现出的则是更为务实的技术路线\n大多以具体场景为导向\n强调技术与实体经济的结合\n这种差异反映了中美产业文化的根本不同\n但是无论哪种模式\n都面临着\"技术万能论\"的陷阱\n当企业将AGI视为一个必须征服的\"圣杯\"，\n就可能忽视技术发展的真实边界和社会成本\nOpenAI的发展历程\n是一部理想主义向商业现实妥协的教科书\n2015年\n埃隆·马斯克（Elon Musk）与萨姆·奥尔特曼（Sam Altman）创立这家非营利组织的时候\n目标是\"通过开放协作防止AI被垄断\"。\n早期的OpenAI确实践行了开源的理念\n将GPT-1的技术细节公之于众\n并且通过\"使命感\"吸引到了顶尖人才\n包括前首席科学家伊利亚·苏茨克维尔（Ilya Sutskever）放弃了谷歌的百万年薪\n正是被\"打造造福全人类的AI\"的愿景所打动\n但是算力的瓶颈改写了一切\n当团队发现深度学习模型的性能提升\n高度依赖于算力规模后\n非营利模式的资金限制就成为了一道致命伤\n2016年，OpenAI决定引入商业投资\n转型为\"营利性公司\"，\n这个决策导致马斯克退出\n他后来公开表示\n\"无法接受将AI技术交给追求利润的资本\"。\n转型后的OpenAI迅速变脸\n停止开源技术\n与微软达成独家云计算协议\n将技术细节作为商业机密\n到2025年\nOpenAI估值已达3000亿美元\n成为全球最值钱的AI企业\n而创立时的\"开放、透明\"原则\n早已被\"规模至上\"的资本逻辑取代\n作为OpenAI的灵魂人物\nSam Altman的崛起堪称硅谷精英的典型样本\n这位YC前总裁的职业生涯\n贯穿了对技术趋势的精准把握和对人性的深刻洞察\n在担任YC总裁期间\n他将这家初创加速器打造成了硅谷的创新引擎\n投资了包括Airbnb、Dropbox在内的明星企业\n积累了深厚的资本人脉\n2015年创立OpenAI时\n他展现出了非凡的八面玲瓏能力\n对风险投资人\n他会强调\"先发优势带来的垄断利润\"；\n对政府官员\n他会大谈\"AI监管的全球协作\"；\n对普通用户\n他则塑造\"技术普惠\"的形象\n这种多面性让外界难以判断他的真实信念\n但是也让他成为资本、技术、政策三界的\"润滑剂\"。\n正如郝珂灵在采访中指出的\nAltman并非传统意义上的技术专家\n而是一名精通\"愿景营销\"的商业领袖\n他的成功也证明\n在硅谷\"讲好技术故事\"，\n有时比技术本身更加重要\n在AI产业的光鲜财报背后\n隐藏着对全球南方劳动力的系统性剥削\nOpenAI的内容审核体系\n依赖于肯尼亚工人的人工标注\n这些工人以每小时2-3美元的报酬\n处理包含暴力、仇恨、色情内容的海量文本\n一名工人曾经在采访中回忆\n他每天需要标注超过500条的极端内容\n三个月后出现了严重的PTSD症状\n无法与家人正常交流\n更令人不安的是\n这些用于训练AI安全系统的数据\n从来没有向工人告知过具体的内容和潜在风险\n导致企业将心理创伤成本完全转嫁给了弱势劳工\n在自动驾驶数据标注领域\n剥削则呈现出了更为隐蔽的形态\n2016年委内瑞拉经济崩溃后\n大量受过高等教育的难民\n在哥伦比亚成为了AI数据标注工人\n他们通过众包平台承接任务\n却陷入了一场\"任务抢夺战\"，\n每个标注任务上线后几秒内就被抢光\n迫使工人24小时保持在线\n一位患有糖尿病的委内瑞拉女性难民\n因为长时间紧盯屏幕导致病情恶化\n最终失明\n而她所标注的自动驾驶数据\n正在被用来开发价值数十亿美元的智能汽车系统\n郝珂灵指出\nAI产业的扩张正在以独特的方式冲击民主制度\n在地方层面\n数据中心的选址过程往往会绕过公众的参与\n比如亚利桑那州的一位市议员\n在批准数据中心建设后才发现\n该项目将消耗当地 30% 的淡水资源\n而此前企业承诺的 \"零影响\" 报告严重失实\n在国家层面\n美国联邦政府通过的《人工智能创新法案》，\n实际上冻结了各州制定本地监管政策的权力\n将监管权集中到了联邦层面\n而联邦政策的制定\n又高度依赖企业提供的技术评估\n从而形成了一种\"监管被俘\"的局面\n更深远的影响还在于社会契约的破坏\n当公众发现自己使用的AI服务\n建立在对全球南方劳工的剥削、对环境法规的践踏之上\n技术合法性将面临根本性的挑战\n郝珂灵在访谈中警告\n如果放任AI巨头的\"帝国式扩张\"，\n未来20年内可能出现所谓的\"技术封建主义\"，\n也就是少数企业会控制关键的基础设施\n公众在数据、算力、服务上形成依赖\n民主制度难以有效制约技术权力\n面对技术集权的威胁\n全球范围内已经涌现出了多元反抗力量\n在智利，社区组织通过三年诉讼\n成功阻止了谷歌在水资源保护区建设耗水型数据中心\n迫使企业采用海水冷却技术；\n在美国\n艺术家集体起诉Stable Diffusion未经授权\n使用120万幅版权作品来训练模型\n推动法院首次明确AI训练数据的版权边界；\n在欧盟\n《人工智能法案》建立了技术风险分级制度\n要求高风险AI系统必须公开训练数据来源\n接受第三方审计\n这些案例证明\n破局的关键在于抓住AI供应链的薄弱环节\n比如数据获取、算力消耗、劳工权益和政策监管\n普通公众可以通过行使数据权利\n比如拒绝非必要的数据收集、支持通过伦理认证的AI产品、参与社区技术听证会等方式\n积少成多形成制约力量\n正如郝珂灵所言\nAI技术的未来不是由少数企业所决定的\n而是取决于我们能否在每个环节\n建立民主参与机制\n确保技术进步服务于公共利益\n而非资本扩张\n九、技术神话的反思\n在采访的最后\n郝珂灵将当前的AI狂热\n与电影《沙丘》中的宗教神话做了深刻类比\n当技术精英不断在强化\nAI将重塑人类文明的二元叙事时\n无论是追捧者还是反对者\n都可能陷入到认知陷阱中\n忽视技术发展的渐进性和复杂性\n历史经验告诉我们\n任何技术革命如果缺乏社会制衡\n都可能异化为新的剥削工具\n正如19世纪的东印度公司\n凭借着技术优势掠夺殖民地资源一样\n21世纪的AI巨头也正在重演类似逻辑\n只不过剥削对象从土地变为数据、算力和人类的注意力\n好了\n以上就是郝珂灵这次访谈的主要内容了\n欢迎大家在评论区发表自己的看法\n感谢观看本期视频，我们下期再见", "language": "zh", "duration": 630.0, "segments": [ { "id": 0, "start": 0.3, "end": 2.766, "text": "大家好，这里是最佳拍档，我是大飞", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 1, "start": 2.966, "end": 6.466, "text": "当我们每天使用AI生成文案、图片甚至视频的时候", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 2, "start": 6.566, "end": 9.566, "text": "是否想过，这些看似神奇的技术背后", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 3, "start": 9.633, "end": 13.5, "text": "正运行着一个由数据、算力和资本构成的庞大帝国", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 4, "start": 13.7, "end": 14.333, "text": "今天", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 5, "start": 14.333, "end": 20.233, "text": "我们将回顾《华尔街日报》科技记者郝珂灵（Karen Hao）在Novara Media的一场专访", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 6, "start": 20.633, "end": 24.166, "text": "看看她是如何通过对硅谷AI行业的深度调查", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 7, "start": 24.3, "end": 28.566, "text": "揭开那些推动技术进步的关键力量如何重塑世界", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 8, "start": 28.666, "end": 31.833, "text": "以及这场技术革命背后隐藏的系统性风险", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 9, "start": 32.3, "end": 35.633, "text": "这期视频我们将从概念的起源到资源的争夺", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 10, "start": 35.733, "end": 38.5, "text": "从企业的蜕变到全球供应链的暗角", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 11, "start": 38.666, "end": 41.033, "text": "深入了解这个正在成型的\"AI帝国\"，", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 12, "start": 41.033, "end": 42.233, "text": "究竟是如何运作的", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 13, "start": 43.133, "end": 45.766, "text": "要想理解当前AI产业的疯狂扩张", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 14, "start": 46.3, "end": 49.4, "text": "必须先厘清\"人工智能\"这个充满模糊的概念", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 15, "start": 49.633, "end": 50.5, "text": "1956年", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 16, "start": 50.5, "end": 53.1, "text": "达特茅斯学院的助理教授约翰·麦卡锡（John McCarthy）", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 17, "start": 53.1, "end": 55.4, "text": "为了给夏季研究项目筹集资金", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 18, "start": 55.566, "end": 57.833, "text": "首次创造了\"人工智能\"这个术语", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 19, "start": 57.966, "end": 60.066, "text": "这个诞生于学术营销的词汇", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 20, "start": 60.333, "end": 64.0, "text": "从此成为了一个可以装入任何技术的\"手提箱概念\"。", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 21, "start": 64.233, "end": 67.1, "text": "如今OpenAI、谷歌等巨头所谈论的AI", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 22, "start": 67.266, "end": 70.1, "text": "本质上是机器学习的一个子类，即深度学习", "source": "video_source_subtitle", "authority": "video_source_subtitle", "text_original": "本质上是机器学习的一个子类深度学习", "correction_applied": true, "correction_source": "llm_correct_a_presegmentation" }, { "id": 23, "start": 70.266, "end": 73.966, "text": "也就是通过神经网络对海量的数据进行模式计算", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 24, "start": 74.033, "end": 76.766, "text": "实现文本生成、图像识别等功能", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 25, "start": 76.966, "end": 80.666, "text": "但是公众很容易将它等同于科幻作品中的通用智能", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 26, "start": 80.766, "end": 83.5, "text": "忽视了它作为一种特定技术工具的本质", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 27, "start": 83.966, "end": 85.3, "text": "这种概念上的模糊", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 28, "start": 85.3, "end": 87.2, "text": "导致了严重的认知偏差", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 29, "start": 87.333, "end": 89.166, "text": "比如当英国政府宣称", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 30, "start": 89.233, "end": 93.566, "text": "用AI为英国国家医疗服务体系节省了20%开支\"时", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 31, "start": 93.866, "end": 96.433, "text": "并没有说明具体用的是哪种AI技术", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 32, "start": 96.633, "end": 99.033, "text": "究竟是基于规则的医疗影像分析", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 33, "start": 99.033, "end": 101.466, "text": "还是需要海量数据训练的大语言模型？", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 34, "start": 101.866, "end": 105.3, "text": "两种技术在成本、效果和实施难度上的差异巨大", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 35, "start": 105.366, "end": 109.166, "text": "但是\"人工智能\"这个笼统的表述掩盖了所有的细节", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 36, "start": 109.333, "end": 110.166, "text": "更关键的是", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 37, "start": 110.166, "end": 113.633, "text": "这种模糊性让公众难以评估技术的真实影响", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 38, "start": 113.733, "end": 115.6, "text": "比如很少有人意识到", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 39, "start": 115.833, "end": 117.933, "text": "训练一个大语言模型的碳排放量", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 40, "start": 117.933, "end": 120.366, "text": "相当于数十万辆汽车的终身排放量", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 41, "start": 121.266, "end": 122.733, "text": "AI技术的每一次突破", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 42, "start": 122.766, "end": 125.2, "text": "都伴随着算力需求的指数级增长", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 43, "start": 125.466, "end": 126.8, "text": "麦肯锡报告显示", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 44, "start": 126.866, "end": 129.966, "text": "未来五年全球AI数据中心的能源消耗", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 45, "start": 130.0, "end": 134.0, "text": "将每年新增相当于英国0.5到1.2倍的用电量", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 46, "start": 134.2, "end": 138.5, "text": "而其中60%以上要依赖煤炭、天然气等化石燃料", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 47, "start": 138.633, "end": 139.633, "text": "在得克萨斯州", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 48, "start": 139.633, "end": 142.8, "text": "多家本应退役的燃煤电厂被重新启用", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 49, "start": 143.0, "end": 146.0, "text": "唯一的目的就是为AI训练提供稳定的电力；", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 50, "start": 146.366, "end": 151.666, "text": "埃隆·马斯克（Elon Musk）的XAI公司在田纳西州孟菲斯建设的Colossus超级计算机", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 51, "start": 151.833, "end": 155.0, "text": "未经许可便安装了35台甲烷燃气轮机", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 52, "start": 155.166, "end": 159.5, "text": "这些设备每天排放的有毒污染物超过了当地工业标准的200%，", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 53, "start": 159.833, "end": 162.0, "text": "严重威胁到周边居民的健康", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 54, "start": 162.266, "end": 163.966, "text": "水资源的争夺同样激烈", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 55, "start": 164.066, "end": 166.666, "text": "由于数据中心需要纯净水冷却设备", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 56, "start": 166.733, "end": 169.2, "text": "全球67%的新建AI数据中心", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 57, "start": 169.466, "end": 171.9, "text": "选址在干旱或者半干旱地区", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 58, "start": 172.166, "end": 175.5, "text": "乌拉圭首都蒙得维的亚正在经历百年一遇的干旱", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 59, "start": 175.666, "end": 179.3, "text": "当地政府不得不将处理后的工业废水混入饮用水", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 60, "start": 179.466, "end": 182.733, "text": "但是谷歌仍计划在此建设新的数据中心", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 61, "start": 183.0, "end": 185.8, "text": "该项目预计每天将消耗8万吨淡水", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 62, "start": 185.866, "end": 188.133, "text": "相当于16万居民的日常用水量", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 63, "start": 188.466, "end": 190.433, "text": "这种\"技术发展优先\"的逻辑", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 64, "start": 190.433, "end": 193.933, "text": "正在将环境成本转嫁到公共资源和弱势群体身上", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 65, "start": 194.833, "end": 197.533, "text": "郝珂灵指出，当前美国AI领域的竞争", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 66, "start": 197.533, "end": 201.266, "text": "本质上是一场关于\"通用人工智能（AGI）\"的集体狂欢", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 67, "start": 201.6, "end": 205.466, "text": "OpenAI、Anthropic、谷歌、Meta等公司投入数百亿美元", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 68, "start": 205.6, "end": 208.266, "text": "声称要创造\"重塑人类智能\"的技术", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 69, "start": 208.566, "end": 210.3, "text": "但是颇具讽刺意味的是", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 70, "start": 210.466, "end": 213.7, "text": "这些公司的商业案例普遍都是模糊不清的", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 71, "start": 214.0, "end": 217.966, "text": "比如微软在2024年悄悄撤回了部分数据中心的投资", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 72, "start": 218.133, "end": 220.133, "text": "承认\"短期回报不及预期\"，", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 73, "start": 220.4, "end": 223.666, "text": "而OpenAI向投资者描绘的\"AGI垄断\"愿景", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 74, "start": 223.833, "end": 227.466, "text": "更多是基于彼得·蒂尔（Peter Thiel）「竞争是给白痴准备的，创业必须要垄断」的理论", "source": "video_source_subtitle", "authority": "video_source_subtitle", "text_original": "更多是基于彼得·蒂尔（Peter Thiel）\"竞争是给白痴准备的", "correction_applied": true, "correction_source": "llm_correct_a_presegmentation" }, { "id": 75, "start": 227.466, "end": 229.266, "text": "创业必须要垄断\"的理论", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 76, "start": 229.433, "end": 231.133, "text": "而非实际的技术路径", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 77, "start": 231.633, "end": 233.266, "text": "与美国的概念驱动不同", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 78, "start": 233.4, "end": 236.733, "text": "中国AI企业展现出的则是更为务实的技术路线", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 79, "start": 236.866, "end": 238.933, "text": "大多以具体场景为导向", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 80, "start": 239.033, "end": 241.166, "text": "强调技术与实体经济的结合", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 81, "start": 241.433, "end": 244.266, "text": "这种差异反映了中美产业文化的根本不同", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 82, "start": 244.4, "end": 245.866, "text": "但是无论哪种模式", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 83, "start": 245.933, "end": 248.066, "text": "都面临着\"技术万能论\"的陷阱", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 84, "start": 248.233, "end": 251.066, "text": "当企业将AGI视为一个必须征服的\"圣杯\"，", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 85, "start": 251.133, "end": 254.4, "text": "就可能忽视技术发展的真实边界和社会成本", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 86, "start": 255.3, "end": 256.3, "text": "OpenAI的发展历程", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 87, "start": 256.3, "end": 259.666, "text": "是一部理想主义向商业现实妥协的教科书", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 88, "start": 259.9, "end": 260.666, "text": "2015年", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 89, "start": 260.666, "end": 263.933, "text": "埃隆·马斯克（Elon Musk）与萨姆·奥尔特曼（Sam Altman）创立这家非营利组织的时候", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 90, "start": 264.166, "end": 267.333, "text": "目标是\"通过开放协作防止AI被垄断\"。", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 91, "start": 267.733, "end": 270.3, "text": "早期的OpenAI确实践行了开源的理念", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 92, "start": 270.4, "end": 272.6, "text": "将GPT-1的技术细节公之于众", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 93, "start": 272.6, "end": 275.333, "text": "并且通过\"使命感\"吸引到了顶尖人才", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 94, "start": 275.533, "end": 279.9, "text": "包括前首席科学家伊利亚·苏茨克维尔（Ilya Sutskever）放弃了谷歌的百万年薪", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 95, "start": 280.0, "end": 283.1, "text": "正是被\"打造造福全人类的AI\"的愿景所打动", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 96, "start": 283.466, "end": 285.433, "text": "但是算力的瓶颈改写了一切", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 97, "start": 285.6, "end": 288.3, "text": "当团队发现深度学习模型的性能提升", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 98, "start": 288.333, "end": 290.2, "text": "高度依赖于算力规模后", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 99, "start": 290.333, "end": 293.2, "text": "非营利模式的资金限制就成为了一道致命伤", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 100, "start": 293.533, "end": 296.033, "text": "2016年，OpenAI决定引入商业投资", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 101, "start": 296.066, "end": 297.533, "text": "转型为\"营利性公司\"，", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 102, "start": 297.766, "end": 299.666, "text": "这个决策导致马斯克退出", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 103, "start": 299.966, "end": 301.2, "text": "他后来公开表示", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 104, "start": 301.3, "end": 304.4, "text": "\"无法接受将AI技术交给追求利润的资本\"。", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 105, "start": 304.733, "end": 306.5, "text": "转型后的OpenAI迅速变脸", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 106, "start": 306.5, "end": 307.6, "text": "停止开源技术", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 107, "start": 307.6, "end": 309.733, "text": "与微软达成独家云计算协议", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 108, "start": 309.766, "end": 311.8, "text": "将技术细节作为商业机密", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 109, "start": 312.2, "end": 313.133, "text": "到2025年", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 110, "start": 313.133, "end": 315.4, "text": "OpenAI估值已达3000亿美元", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 111, "start": 315.466, "end": 317.6, "text": "成为全球最值钱的AI企业", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 112, "start": 317.766, "end": 320.233, "text": "而创立时的\"开放、透明\"原则", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 113, "start": 320.366, "end": 322.633, "text": "早已被\"规模至上\"的资本逻辑取代", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 114, "start": 323.533, "end": 324.966, "text": "作为OpenAI的灵魂人物", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 115, "start": 325.0, "end": 328.033, "text": "Sam Altman的崛起堪称硅谷精英的典型样本", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 116, "start": 328.2, "end": 330.133, "text": "这位YC前总裁的职业生涯", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 117, "start": 330.2, "end": 334.066, "text": "贯穿了对技术趋势的精准把握和对人性的深刻洞察", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 118, "start": 334.366, "end": 335.966, "text": "在担任YC总裁期间", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 119, "start": 335.966, "end": 339.333, "text": "他将这家初创加速器打造成了硅谷的创新引擎", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 120, "start": 339.433, "end": 342.733, "text": "投资了包括Airbnb、Dropbox在内的明星企业", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 121, "start": 342.8, "end": 344.566, "text": "积累了深厚的资本人脉", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 122, "start": 344.833, "end": 346.533, "text": "2015年创立OpenAI时", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 123, "start": 346.6, "end": 349.1, "text": "他展现出了非凡的八面玲瓏能力", "source": "video_source_subtitle", "authority": "video_source_subtitle", "text_original": "他展现出了非凡的千人千面能力", "correction_applied": true, "correction_source": "llm_correct_a_presegmentation" }, { "id": 124, "start": 349.233, "end": 350.233, "text": "对风险投资人", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 125, "start": 350.233, "end": 353.166, "text": "他会强调\"先发优势带来的垄断利润\"；", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 126, "start": 353.333, "end": 354.166, "text": "对政府官员", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 127, "start": 354.166, "end": 356.5, "text": "他会大谈\"AI监管的全球协作\"；", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 128, "start": 356.7, "end": 357.5, "text": "对普通用户", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 129, "start": 357.5, "end": 359.566, "text": "他则塑造\"技术普惠\"的形象", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 130, "start": 359.833, "end": 362.966, "text": "这种多面性让外界难以判断他的真实信念", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 131, "start": 363.1, "end": 366.766, "text": "但是也让他成为资本、技术、政策三界的\"润滑剂\"。", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 132, "start": 367.033, "end": 369.033, "text": "正如郝珂灵在采访中指出的", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 133, "start": 369.233, "end": 371.566, "text": "Altman并非传统意义上的技术专家", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 134, "start": 371.633, "end": 374.366, "text": "而是一名精通\"愿景营销\"的商业领袖", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 135, "start": 374.7, "end": 375.833, "text": "他的成功也证明", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 136, "start": 375.833, "end": 377.733, "text": "在硅谷\"讲好技术故事\"，", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 137, "start": 377.733, "end": 379.766, "text": "有时比技术本身更加重要", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 138, "start": 380.666, "end": 382.7, "text": "在AI产业的光鲜财报背后", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 139, "start": 382.8, "end": 385.833, "text": "隐藏着对全球南方劳动力的系统性剥削", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 140, "start": 386.133, "end": 387.766, "text": "OpenAI的内容审核体系", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 141, "start": 387.766, "end": 389.833, "text": "依赖于肯尼亚工人的人工标注", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 142, "start": 389.966, "end": 392.7, "text": "这些工人以每小时2-3美元的报酬", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 143, "start": 392.766, "end": 396.133, "text": "处理包含暴力、仇恨、色情内容的海量文本", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 144, "start": 396.4, "end": 398.433, "text": "一名工人曾经在采访中回忆", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 145, "start": 398.5, "end": 401.266, "text": "他每天需要标注超过500条的极端内容", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 146, "start": 401.333, "end": 404.066, "text": "三个月后出现了严重的PTSD症状", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 147, "start": 404.166, "end": 405.866, "text": "无法与家人正常交流", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 148, "start": 406.166, "end": 407.2, "text": "更令人不安的是", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 149, "start": 407.2, "end": 409.7, "text": "这些用于训练AI安全系统的数据", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 150, "start": 409.766, "end": 413.133, "text": "从来没有向工人告知过具体的内容和潜在风险", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 151, "start": 413.2, "end": 416.833, "text": "导致企业将心理创伤成本完全转嫁给了弱势劳工", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 152, "start": 417.133, "end": 418.866, "text": "在自动驾驶数据标注领域", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 153, "start": 418.9, "end": 421.4, "text": "剥削则呈现出了更为隐蔽的形态", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 154, "start": 421.733, "end": 424.366, "text": "2016年委内瑞拉经济崩溃后", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 155, "start": 424.366, "end": 426.2, "text": "大量受过高等教育的难民", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 156, "start": 426.266, "end": 428.666, "text": "在哥伦比亚成为了AI数据标注工人", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 157, "start": 428.8, "end": 430.766, "text": "他们通过众包平台承接任务", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 158, "start": 430.766, "end": 432.633, "text": "却陷入了一场\"任务抢夺战\"，", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 159, "start": 432.9, "end": 435.4, "text": "每个标注任务上线后几秒内就被抢光", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 160, "start": 435.6, "end": 437.866, "text": "迫使工人24小时保持在线", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 161, "start": 438.1, "end": 440.7, "text": "一位患有糖尿病的委内瑞拉女性难民", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 162, "start": 440.733, "end": 443.4, "text": "因为长时间紧盯屏幕导致病情恶化", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 163, "start": 443.4, "end": 444.266, "text": "最终失明", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 164, "start": 444.366, "end": 446.266, "text": "而她所标注的自动驾驶数据", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 165, "start": 446.3, "end": 449.7, "text": "正在被用来开发价值数十亿美元的智能汽车系统", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 166, "start": 450.6, "end": 451.566, "text": "郝珂灵指出", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 167, "start": 451.566, "end": 455.1, "text": "AI产业的扩张正在以独特的方式冲击民主制度", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 168, "start": 455.366, "end": 456.333, "text": "在地方层面", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 169, "start": 456.366, "end": 459.4, "text": "数据中心的选址过程往往会绕过公众的参与", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 170, "start": 459.4, "end": 461.5, "text": "比如亚利桑那州的一位市议员", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 171, "start": 461.6, "end": 464.033, "text": "在批准数据中心建设后才发现", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 172, "start": 464.2, "end": 467.333, "text": "该项目将消耗当地 30% 的淡水资源", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 173, "start": 467.533, "end": 470.8, "text": "而此前企业承诺的 \"零影响\" 报告严重失实", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 174, "start": 471.0, "end": 471.933, "text": "在国家层面", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 175, "start": 471.933, "end": 474.733, "text": "美国联邦政府通过的《人工智能创新法案》，", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 176, "start": 474.866, "end": 478.3, "text": "实际上冻结了各州制定本地监管政策的权力", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 177, "start": 478.533, "end": 480.633, "text": "将监管权集中到了联邦层面", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 178, "start": 480.866, "end": 482.1, "text": "而联邦政策的制定", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 179, "start": 482.1, "end": 484.466, "text": "又高度依赖企业提供的技术评估", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 180, "start": 484.466, "end": 486.733, "text": "从而形成了一种\"监管被俘\"的局面", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 181, "start": 487.2, "end": 489.833, "text": "更深远的影响还在于社会契约的破坏", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 182, "start": 490.066, "end": 492.333, "text": "当公众发现自己使用的AI服务", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 183, "start": 492.333, "end": 496.9, "text": "建立在对全球南方劳工的剥削、对环境法规的践踏之上", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 184, "start": 497.066, "end": 499.533, "text": "技术合法性将面临根本性的挑战", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 185, "start": 499.933, "end": 501.533, "text": "郝珂灵在访谈中警告", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 186, "start": 501.666, "end": 504.0, "text": "如果放任AI巨头的\"帝国式扩张\"，", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 187, "start": 504.133, "end": 507.133, "text": "未来20年内可能出现所谓的\"技术封建主义\"，", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 188, "start": 507.2, "end": 510.233, "text": "也就是少数企业会控制关键的基础设施", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 189, "start": 510.4, "end": 513.233, "text": "公众在数据、算力、服务上形成依赖", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 190, "start": 513.366, "end": 515.6, "text": "民主制度难以有效制约技术权力", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 191, "start": 516.5, "end": 518.033, "text": "面对技术集权的威胁", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 192, "start": 518.166, "end": 521.033, "text": "全球范围内已经涌现出了多元反抗力量", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 193, "start": 521.3, "end": 524.033, "text": "在智利，社区组织通过三年诉讼", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 194, "start": 524.066, "end": 528.033, "text": "成功阻止了谷歌在水资源保护区建设耗水型数据中心", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 195, "start": 528.266, "end": 530.4, "text": "迫使企业采用海水冷却技术；", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 196, "start": 530.6, "end": 531.366, "text": "在美国", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 197, "start": 531.366, "end": 534.5, "text": "艺术家集体起诉Stable Diffusion未经授权", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 198, "start": 534.6, "end": 537.7, "text": "使用120万幅版权作品来训练模型", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 199, "start": 537.933, "end": 541.433, "text": "推动法院首次明确AI训练数据的版权边界；", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 200, "start": 541.766, "end": 542.5, "text": "在欧盟", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 201, "start": 542.5, "end": 545.3, "text": "《人工智能法案》建立了技术风险分级制度", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 202, "start": 545.466, "end": 548.9, "text": "要求高风险AI系统必须公开训练数据来源", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 203, "start": 548.9, "end": 550.333, "text": "接受第三方审计", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 204, "start": 550.8, "end": 551.866, "text": "这些案例证明", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 205, "start": 551.9, "end": 555.066, "text": "破局的关键在于抓住AI供应链的薄弱环节", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 206, "start": 555.133, "end": 559.2, "text": "比如数据获取、算力消耗、劳工权益和政策监管", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 207, "start": 559.6, "end": 561.933, "text": "普通公众可以通过行使数据权利", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 208, "start": 561.933, "end": 569.033, "text": "比如拒绝非必要的数据收集、支持通过伦理认证的AI产品、参与社区技术听证会等方式", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 209, "start": 569.266, "end": 571.233, "text": "积少成多形成制约力量", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 210, "start": 571.466, "end": 572.7, "text": "正如郝珂灵所言", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 211, "start": 572.933, "end": 575.666, "text": "AI技术的未来不是由少数企业所决定的", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 212, "start": 575.666, "end": 578.133, "text": "而是取决于我们能否在每个环节", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 213, "start": 578.133, "end": 579.7, "text": "建立民主参与机制", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 214, "start": 579.8, "end": 581.9, "text": "确保技术进步服务于公共利益", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 215, "start": 581.9, "end": 582.966, "text": "而非资本扩张", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 216, "start": 582.966, "end": 583.7, "text": "九、技术神话的反思", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 217, "start": 583.866, "end": 584.9, "text": "在采访的最后", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 218, "start": 584.9, "end": 586.833, "text": "郝珂灵将当前的AI狂热", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 219, "start": 586.833, "end": 590.2, "text": "与电影《沙丘》中的宗教神话做了深刻类比", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 220, "start": 590.366, "end": 592.233, "text": "当技术精英不断在强化", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 221, "start": 592.366, "end": 594.9, "text": "AI将重塑人类文明的二元叙事时", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 222, "start": 595.033, "end": 597.066, "text": "无论是追捧者还是反对者", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 223, "start": 597.066, "end": 599.166, "text": "都可能陷入到认知陷阱中", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 224, "start": 599.266, "end": 601.966, "text": "忽视技术发展的渐进性和复杂性", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 225, "start": 602.233, "end": 603.5, "text": "历史经验告诉我们", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 226, "start": 603.566, "end": 606.1, "text": "任何技术革命如果缺乏社会制衡", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 227, "start": 606.1, "end": 608.566, "text": "都可能异化为新的剥削工具", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 228, "start": 608.666, "end": 610.8, "text": "正如19世纪的东印度公司", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 229, "start": 610.866, "end": 613.6, "text": "凭借着技术优势掠夺殖民地资源一样", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 230, "start": 613.9, "end": 617.2, "text": "21世纪的AI巨头也正在重演类似逻辑", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 231, "start": 617.3, "end": 622.133, "text": "只不过剥削对象从土地变为数据、算力和人类的注意力", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 232, "start": 622.366, "end": 622.9, "text": "好了", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 233, "start": 622.9, "end": 625.266, "text": "以上就是郝珂灵这次访谈的主要内容了", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 234, "start": 625.266, "end": 627.233, "text": "欢迎大家在评论区发表自己的看法", "source": "video_source_subtitle", "authority": "video_source_subtitle" }, { "id": 235, "start": 627.266, "end": 629.5, "text": "感谢观看本期视频，我们下期再见", "source": "video_source_subtitle", "authority": "video_source_subtitle" } ], "transcript_selection": { "schema": "videonote.transcript_selection.v1", "status": "SELECTED", "requested_language": "zh", "duration": 630.0, "reference_path": "/Users/sagawa/AI_WORK/video_notes/url/20260907_085812__人工智能_AI帝国的背后真相_Karen_Hao_华尔街日报_记者_人工智能概念被滥用_电力与环境_AGI的集体狂欢_OpenAI的转向_AI底层劳工_技术神话/reference_subtitle.vtt", "reference_kind": "manual", "reference_language": "zh-Hans", "selected_source": "video_source_subtitle", "source_caption_kind": "manual", "source_caption_processing": "verbatim_direct", "youtube_overlap_clamps": [], "general_high_confidence_repairs": [], "decision_reason": "complete_source_subtitle_fast_path_no_whisper", "whisper_rejected": false, "source_profile": { "segment_count": 236, "duration_coverage": 0.935202 } }, "presegmentation_correction": { "applied": true, "correction_count": 3, "deterministic_fix_count": 0 } }