DeepMind的年度报告：亏损42亿再创新高，收益仰赖谷歌母公司

{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"italic"},{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"本文是我们探索"},{"type":"link","attrs":{"href":"https:\/\/bdtechtalks.com\/tag\/ai-business\/","title":null,"type":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"人工智能商业"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"italic"},{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"系列的一部分"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"上周，DeepMind在利用人工智能预测蛋白质折叠方面取得突破性进展，紧随其后有消息称，这家总部位于英国的AI公司仍在让其母公司Alphabet承受每年数亿美元的亏损。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"高科技公司亏损并不是什么新鲜事了。科技行业中充斥着许多公司的例子，这些公司远在盈利之前就已经烧掉了投资者的钱。但是DeepMind并不是一家寻求抢占特定市场份额的普通公司，它是一个AI研究实验室，但不得不将自己重新定位为半商业机构，以确保其生存。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"尽管其所有者（也是谷歌的母公司）目前乐意支付DeepMind昂贵的AI研究费用，但并不能保证这将永远持续下去。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"heading","attrs":{"align":null,"level":1},"content":[{"type":"text","text":"DeepMind的盈亏"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.geekbang.org\/infoq\/d0\/d07acb46eec6481d3dae4e21fcde25ba.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"DeepMind的AlphaFold项目利用人工智能来帮助应对复杂的蛋白质折叠挑战"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"根据向英国公司注册处提交的"},{"type":"link","attrs":{"href":"https:\/\/find-and-update.company-information.service.gov.uk\/company\/07386350\/filing-history\/MzI4NTg2NzI0NWFkaXF6a2N4\/document?format=pdf&download=0","title":null,"type":null},"content":[{"type":"text","text":"年度报告"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"，DeepMind的收入翻了一倍还多，从2018年的1.03亿英镑增长到2019年的2.66亿英镑。但公司的支出也持续增长，从 2018年的5.68亿英镑增长到2019年的7.17亿英镑。公司的整体亏损从2018年的4.7亿英镑增加到2019年的4.77亿英镑（约合42亿人民币）。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"乍看之下，这不是坏消息。"},{"type":"link","attrs":{"href":"https:\/\/bdtechtalks.com\/2019\/08\/12\/deepmind-losses-costs-of-ai\/","title":null,"type":null},"content":[{"type":"text","text":"与前几年相比"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"，DeepMind的收入增长正在加速，同时它的亏损却保持稳定。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.geekbang.org\/infoq\/f1\/f18c6bf293e5ccd5f4544bf71b5dca12.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"DeepMind从2016年到2019年的收入和亏损"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"但是该报告包含了一些更重要的事实。该文档提到“研发报酬来自于集团其他事业部”。这意味着DeepMind的主要客户是它的所有者。 Alphabet正在向DeepMind支付费用，以将其AI研究和人才应用于谷歌的服务和基础架构。过去，谷歌使用DeepMind的服务来完成诸如管理其数据中心的电网以及改善其语音助手的人工智能之类的任务。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"这也意味着DeepMind的AI尚无市场，如果有，也只能通过谷歌获得。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"该文档还提到，成本的增长“主要与技术基础设施，员工成本和其他相关费用的增长有关。”"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"这是很重要的一点。 DeepMind的“技术基础架构”主要运行在谷歌的大型云计算服务及其特殊的AI处理器Tensor Processing Unit（TPU）上。 DeepMind的主要研究领域是"},{"type":"link","attrs":{"href":"https:\/\/bdtechtalks.com\/2019\/05\/28\/what-is-reinforcement-learning\/","title":null,"type":null},"content":[{"type":"text","text":"深度强化学习"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"，这需要访问非常昂贵的计算资源。该公司在2019年的一些项目中包括了两套AI系统的研究工作，其中一个系统可以玩《"},{"type":"link","attrs":{"href":"https:\/\/bdtechtalks.com\/2019\/11\/04\/deepmind-ai-starcraft-2-reinforcement-learning\/","title":null,"type":null},"content":[{"type":"text","text":"星际争霸2"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"》，另一个可以玩《"},{"type":"link","attrs":{"href":"https:\/\/bdtechtalks.com\/2019\/06\/03\/deepmind-ai-quake-iii-arena-ctf\/","title":null,"type":null},"content":[{"type":"text","text":"雷神之锤3"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"》，这两个项目的训练费用均为数百万美元。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"DeepMind的一位发言人对媒体说，文档中提到的费用还包括在"},{"type":"link","attrs":{"href":"https:\/\/deepmind.com\/blog\/article\/alphafold-a-solution-to-a-50-year-old-grand-challenge-in-biology","title":null,"type":null},"content":[{"type":"text","text":"AlphaFold"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"项目上的工作，AlphaFold是该公司另一个非常昂贵的著名的蛋白质折叠AI项目。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.geekbang.org\/infoq\/2c\/2ce1ae60526cc5912ac2c8010f4272f1.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"关于谷歌向DeepMind收取多少访问其云AI服务的费用，目前没有公开细节，但谷歌很可能以折扣价出租其TPU。这意味着，如果没有谷歌的支持，该公司的支出将会高得多。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"员工成本是另一个重要的问题。尽管在过去几年中参与学习机器学习课程的人数有所增加，但能够参与DeepMind所从事的尖端AI研究的科学家却非常稀少。从某些方面来看，顶尖的AI人才的"},{"type":"link","attrs":{"href":"https:\/\/www.nytimes.com\/2018\/04\/19\/technology\/artificial-intelligence-salaries-openai.html","title":null,"type":null},"content":[{"type":"text","text":"薪水为七位数"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"对"},{"type":"link","attrs":{"href":"https:\/\/bdtechtalks.com\/2019\/02\/15\/what-is-deep-learning-neural-networks\/","title":null,"type":null},"content":[{"type":"text","text":"深度学习"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"及其在商业环境中的适用性的兴趣与日俱增，这在科技公司之间引起了争夺顶尖AI人才的竞赛。业内大多数顶尖的AI科学家和先驱在谷歌、Facebook、亚马逊和微软等大型公司从事全职或半职工作。抢夺顶尖AI人才的激烈竞争产生了两种后果。首先，就像其他任何供不应求的领域一样，这导致了AI科学家的薪水急剧上升。其次，它导致许多AI科学家从无法负担巨额薪水的学术研究机构转向了有支付能力的富裕科技公司。为了继续进行科学研究，一些科学家继续留在学术界，但是他们的数量太少了，相距甚远。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"如果没有像谷歌这样的大型科技公司的支持，DeepMind这样的研究实验室就无法为他们的项目雇用新的研究人员。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"因此，尽管DeepMind表现出缓慢扭转亏损局面的迹象，但其增长使其更加依赖谷歌的财务资源和大型云基础架构。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"heading","attrs":{"align":null,"level":1},"content":[{"type":"text","text":"谷歌仍然对DeepMind感到满意"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.geekbang.org\/infoq\/36\/36065fc7294f49210ca18ae98bfd5e5e.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"DeepMind开发了一种称为AlphaStar的AI系统，可以在实时策略游戏《星际争霸2》中击败最佳玩家"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"根据DeepMind的年度报告，Alphabet的投资分支之一谷歌 Ireland无限控股公司，“免除了偿还公司间贷款和所有累计利息，总计11亿英镑”"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"DeepMind还获得了谷歌的书面保证，称其将“在至少十二个月的时间里”向该AI公司“继续提供充足的财务支持”。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"目前，谷歌似乎对DeepMind取得的进展感到满意，这也反映在谷歌和Alphabet高管的讲话中。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"在7月的针对投资者和分析师的季度财报电话会议上，Alphabet首席执行官Sundar Pichai说：“我对我们研发团队在AI上的发展节奏感到非常满意。对于我来说重要的是，我们是一家最先进的公司，并且我们处于领先地位。对于我来说，我对我们在谷歌和DeepMind的工程和研发团队的工作节奏感到兴奋。”"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"但是企业界和科学研究的步伐是不同的。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"科学研究以数十年为时间单位来度量。当今在商业应用中使用的许多AI技术从1970年代和1980年代就在开发中了。同样，今天在AI会议上展示的许多前沿研究和技术可能在未来几年内都不会进入大众市场。根据最乐观的估计，DeepMind的最终目标——开发"},{"type":"link","attrs":{"href":"https:\/\/bdtechtalks.com\/2020\/05\/13\/what-is-artificial-general-intelligence-agi\/","title":null,"type":null},"content":[{"type":"text","text":"通用人工智能"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"（AGI）——至少需要数十年的时间。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"另一方面，股东和投资者们的耐心以数月和数年为单位度量。无法在数年内实现盈利，或至少显示出增长迹象希望的公司，会给投资者带来困扰。DeepMind目前哪个都没有，它没有可衡量的增长，因为它唯一的客户只有谷歌本身。而且目前尚不清楚它的一些技术何时（如果有的话）可以商业化。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.geekbang.org\/infoq\/e2\/e29cc8331a80b64288ba6fc0537f707e.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"谷歌首席执行官Sundar Pichai对DeepMind的AI研发进度感到满意。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"这就是DeepMind的困境所在。从本质上讲，它是一个研究实验室，希望突破科学发展的极限，并确保AI的进步对人类有益。但是，其所有者的目标是开发能够解决特定问题并实现盈利的产品。这两个目标是截然相反的，将DeepMind拉向两个不同的方向：保持其科研性质或转变为一家制造产品的AI公司。过去，该公司在寻求科学研究和产品开发的平衡方面"},{"type":"link","attrs":{"href":"https:\/\/bdtechtalks.com\/2019\/08\/26\/deepmind-mustafa-suleyman-commercial-ai\/","title":null,"type":null},"content":[{"type":"text","text":"已经遇到了麻烦"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"然而，DeepMind并不孤单。 DeepMind的隐性竞争对手OpenAI面临着"},{"type":"link","attrs":{"href":"https:\/\/bdtechtalks.com\/2020\/08\/17\/openai-gpt-3-commercial-ai\/","title":null,"type":null},"content":[{"type":"text","text":"类似的身份危机"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"，从AI研究实验室转变为由微软支持的营利性公司，后者"},{"type":"link","attrs":{"href":"https:\/\/bdtechtalks.com\/2020\/09\/21\/gpt-3-economy-business-model\/","title":null,"type":null},"content":[{"type":"text","text":"租用了其深度学习模型"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]},{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"因此，尽管DeepMind还无需为自己无利可图的研究工作而担忧，但随着它受其所有者的影响越来越深，它应该深思自己的未来和AI科学研究的未来。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}},{"type":"strong"}],"text":"原文链接："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"link","attrs":{"href":"https:\/\/bdtechtalks.com\/2020\/12\/21\/deepminds-annual-report-why-its-hard-to-run-a-commercial-ai-lab\/","title":null,"type":null},"content":[{"type":"text","text":"https:\/\/bdtechtalks.com\/2020\/12\/21\/deepminds-annual-report-why-its-hard-to-run-a-commercial-ai-lab\/"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]}]}