AI与可持续竞争优势?

多客科技

作者：微信文章
最近Jay Barney在Sloan Management Review又跟两位同事一块儿写了一篇文章，说这个AI不能带来可持续竞争优势，基本上是重复他跟Martin Reeves（2024，HBR）的那个文章的观点。这次说的更加的斩钉截铁，更加的绝对。有一些想法。还是想列出来跟大家探讨。下面先介绍的是他的文章的精要，后面是我的3点评论和拓展。

文章要点：

Part of AI’s value is that it is digital, and therefore it is fun­damentally copyable, scalable, repeatable, predictable, and uniform.

AI的部分价值在于其数字化属性，因此本质上具备可复制、可扩展、可重复、可预测且标准化的特征。



Companies must cultivate creativity, drive, and passion. That creativity must be technical, involving research and develop¬ment. It must include novel ways to use AI. But it also includes conceiving of novel partnerships and finding novel ways to con¬nect with customers. These are the same pillars of innovativeness that have always distinguished great companies; AI does not change any of it.

企业必须培育创造力、驱动力与热忱。这种创造力需具备技术属性，涵盖研发环节；既包括AI应用的创新模式，也涉及新型合作关系的构建与客户连接方式的革新。这些正是历来区分卓越企业的创新核心支柱，AI并未改变这一本质。



AI can be at the center of a product, a strat­egy, or even a company’s DNA, but it cannot be at the center of a sustainable competi­tive advantage.

AI可以成为产品、战略乃至企业DNA的核心，但无法构成可持续竞争优势的核心。



Why might a company believe that the benefits of AI will somehow accrue to it but not to its competitors? Some reasons include access to capital or hardware, novel algorithms, advanced models, superior engineering talent, business insight, or pro­prietary data. Dissecting each of these reasons reveals that none is sustainable.

为何企业会认为AI带来的收益将唯独惠及自身而非竞争对手？部分原因包括资金或硬件获取优势、新颖算法、先进模型、顶尖工程人才、商业洞察力或专有数据。但对这些因素逐一剖析后会发现，没有任何一项具备可持续性。（豆包翻译）。

这是我今天即兴想到的一个简单的回应与拓展（原本文为英文。下面中文由豆包提供）：  

1. AI的扩散速度

2. AI的私域化应用

3. 新的异质性来源



第一，巴尼（Barney）忽略的是人工智能技术的扩散速度。最终，技术将趋于同质化，所有企业都会回归平均水平。但这一过程可能耗时极长，甚至或许永远无法达到所有企业在人工智能应用上实现竞争均势的均衡状态——部分企业会更擅长、更敏捷，或在应用深度上更具优势。

第二，如果一家企业更善于将人工智能“本土化应用”或“场景化落地”，即便无法构建可持续竞争优势，至少也能创造出暂时性优势。例如，许多传统耳机制造商在声学技术方面造诣深厚，但在无线时代，连接性（如蓝牙或其他无线模式）成为核心功能之一。尽管几乎所有耳机企业都能公开获取连接技术，但部分企业善于将其与自身卓越的声学技术相融合，而其他企业则在反应速度或应用熟练度上稍逊一筹。因此，一些更擅长运用无线技术的新兴企业得以超越传统厂商，获得竞争优势。

若技术发展路径具有多阶段性且存在路径依赖，那么暂时性优势的持续积累可能转化为综合性的长期可持续优势。

人工智能还可能催生其他新的异质性来源。正如巴尼所主张的，许多技术本身无法创造可持续优势，但我认为可以作为构建可持续优势的载体。诚然，我们不能说电力、800免费电话或电子邮件系统本身能创造竞争优势，但正如我在前两点中所提及的，率先采用或更熟练运用这些技术的企业，终将获得暂时性优势。

以电子邮件为例：美国在线（AOL）收购时代华纳（Time-Warner）后，要求后者使用AOL的电子邮件系统，但时代华纳予以拒绝——原因是该系统不够可靠。即便同属一家集团，子公司也会因技术性能不佳而放弃使用母公司的系统。这充分说明，即便企业能够便捷地获取相同技术，其应用效果也可能存在天壤之别，进而导致优势或劣势的分化。

再举一个例子：若互联网普及后，部分医生能通过网络找到治疗某种疾病的新方法，而其他技术更精湛但无法获取该信息的医生则会处于劣势。互联网上的信息触手可及，但不同使用者的获取能力存在差异。同理，人工智能可将大量医学文献即时翻译成多种语言，医生借助AI学习疾病治疗新方法的速度和广度也会各不相同——这至少会为部分医生创造暂时性的竞争优势。

第三，若用户群体规模足够大，特定人工智能技术的应用还将进一步放大异质性。更多人（尤其是更多精英人才）会探索出更贴合特定用户需求的AI定制化方案；反之，若用户群体规模较小，初始阶段便缺乏足够的多样性基础，异质性也难以形成。总体而言，尽管人工智能本身具有同质化效应，但同时也在不断催生新的异质性。

1 Diffusion Speed of AI

2 Domestication of AI

3 New sources of heterogeneity



What Barney dismissed is the speed of the diffusion of AI technology. Eventually it will homogenize and everyone regresses to the mean to the average. But the process takes very long or never perhaps reaches the equilibrium where everyone is at competitive parity in using AI, some are better, some are quicker, some are more.

A firm more skilled at domesticize or using situated AI will create competitive advantage, at least temporary advantage, if not sustainable advantage. For instance, many traditional headphone/earphone producers are very good at acoustics technology in making headphones or earphones. Yet, in the wireless era, the one of the key features is connectivity, e.g., blue tooth or other wireless mode. And the connectivity tech that is openly available to essentially any headphone companies, but some companies are good at utilizing that and blending them with their acoustic excellence, other firms arenot as quick or skillful. And hence, there will be some new firms that utilizing the wireless better than the traditional ones, gaining their competitive advantage.

If the technology is multi-phased in its trajectory and it is path-dependent, then accumulation of temper advantages could translate into enveloped sustainable long term advantage.

AI could create other sources of new heterogeneity. A lot of technology itself as Barney argued, could not create sustainable advantage, but it could serve as a medium through which sustainable advantage could be based. Sure, you couldn't say electricity creates advantage, sustainable advantage. You couldn't say one eight hundred free numbers create advantage, you couldn't say that email system create advantage. But, as I mentioned in the first two points, whoever adopted this earlier or more skillfully will create temporary advantage.

For instance, take email for example. When Time-Warner were acquired by AOL, AOL wanted T-W to use the AOL email system. And T-W refused because the email was not reliable, so even for the same firm they don't use their parent company’s lousy email system. So suffices it to say that firms access to the same technology even when it is readily available could create different results, either advantage or disadvantage.

Take another example. If the internet is widely available and some doctors could find on the internet a method to cure a particular disease. And if other doctors more skillful than this particular doctor did not have such an access will be at a disadvantage. The information is readily available on the internet, but people's access to the internet will differ. Same logic, AI can translate many medical documents instantly into different languages, doctors could use AI to learn new methods of curing disease and their learning speed and scope will be different, and this will at least create temporary competitive advantage for some.

The adoption of a particular AI Technology will also create more heterogeneity if the sample space is large enough. There will be more people, more smart people find better ways of customizing AI for particular users. If the user population is small and you don't have such variety or diversity to begin with. In general, while AI is a homogenization process, it also creates heterogeneity.