内容来源：https://aibusiness.com/generative-ai/microsoft-frontier-company-humans-scale-ai

内容总结：

谷歌云特约报道：选择你的首个生成式AI应用场景

要入门生成式AI，首先应关注那些能够改善人类信息获取体验的领域。

微软前沿公司（Microsoft Frontier Company）的最新案例再次证明，要实现AI投资回报，专家介入不可或缺。微软于周四宣布成立这一全新业务部门，旨在为企业提供部署和整合AI所需的专业能力。该公司将投资25亿美元，组建一支由6000名行业及工程专家组成的团队，与客户共同设计、共同创新、部署并优化AI系统。该计划不仅是为了加速AI项目落地，也承认了一个事实：没有人类参与，企业AI整合将无法成功。

微软表示，这不仅仅是“前沿部署工程”的投资。这一新举措紧随Palantir等组织将软件工程师嵌入企业的做法，也与本周早些时候AWS宣布的10亿美元FDE投资类似。OpenAI也在今年2月推出了类似项目。

云服务商纷纷推动将人类专家嵌入客户环境，这反映出：尽管AI市场快速增长，但企业在全面扩展和部署AI项目时仍存在诸多障碍。

“很多企业仍在AI应用上挣扎，”Omdia分析师苏连杰表示，“即使采用了AI的企业也难以扩大规模，而成功扩大规模的又在维护或安全运营上遇到困难。”

为弥合这一差距，云服务商正在提供包含“前沿部署工程师”在内的问题解决工具。

苏连杰补充道：“从更宏观的角度看，每一个原本在AI生态中某一环节的公司，要么向上游、要么向下游延伸。”他以前沿模型提供商如今也提供SaaS服务和基础设施，以及云服务商提供集成服务和咨询为例，说明这一趋势。

这一趋势也表明，供应商正在修正此前“AI不需要人类辅助”的立场。

“所有AI公司，今天参与AI的每一个人，都必须回答这个问题：为什么现在我们被告知，需要在组织内部嵌入人员，才能从AI中获得正向ROI？”Futurum Group分析师大卫·尼科尔森指出，“如果AI真的那么好，只需开放使用，人们就会许可、使用并从中获益，那你为什么还要把自己置于必须争抢稀缺人才的境地？”

事实上，仅靠自动化难以实现ROI，这一点从福特等企业的最新动态中可见一斑。据报道，福特在依赖自动化后，正在重新雇佣老员工。

福特的逆转是另一个例证，表明AI的市场策略已从“一切皆可自动化”转向“人的不可或缺”。

尼科尔森评价道：“在我看来，他们是在承认，AI部署的复杂性远超我们此前被引导相信的程度。相关的成本更高，而ROI的实现需要更长时间。”

对于企业而言，最佳做法是先明确自己对AI的愿景，然后决定如何实现——无论是使用微软的服务、选择其他云服务商，还是结合不同供应商和提供商的工具。

“根本而言，企业需要理解他们想通过AI实现什么、如何实现，然后从后往前倒推规划路径。”苏连杰总结道。

中文翻译：

由谷歌云赞助
选择你的首个生成式AI应用场景
要开始使用生成式AI，首先应聚焦于那些能够改善人类与信息交互体验的领域。

微软前沿公司（Microsoft Frontier Company）是最新例证，表明专家对于实现AI投资回报是必不可少的。
微软周四宣布成立一项新的运营业务，旨在为企业提供部署和集成AI所需的专业知识。
该公司将向微软前沿公司投资25亿美元，此举将安排6000名行业和工程专家入驻，与客户共同设计、共同创新、部署和改进AI系统。尽管该项目旨在加速AI应用落地，但也承认在企业中，若无人类参与，AI集成无法成功。

虽然微软表示这不仅仅是对一线部署工程的投资，但这项新举措紧跟Palantir等组织的步伐——后者一直倡导将软件工程师嵌入企业——同时也效仿了AWS本周早些时候10亿美元的FDE投资。OpenAI也在今年二月推出了类似项目。

云服务商推动将人类专业知识嵌入客户环境中，这一举措认识到，尽管AI市场快速增长，但企业中仍存在阻碍其全面扩展和部署AI项目的差距。
Omdia（Informa TechTarget旗下部门）分析师连杰·苏（Lian Jye Su）表示：“许多企业仍在采用AI方面挣扎。即使成功采用，也难以规模化；而实现规模化的企业，在安全维护或运营上同样面临挑战。”

为弥合这一差距，云服务商现在正提供包括FDE在内的问题解决工具。
“从更宏观的角度看，过去在AI生态系统中专注于特定层级的每一家公司，如今都在向上游或下游拓展，”苏补充道。他举例指出，前沿模型提供商现在提供SaaS服务和基础设施，而云服务商则提供集成服务和咨询工作。

这一趋势也表明，供应商正在修正此前“AI无需人类协助”的立场。
Futurum集团分析师大卫·尼科尔森（David Nicholson）表示：“所有这些AI公司，以及今天任何与AI相关的参与者，都必须回答这个问题：为什么我们现在被告知，需要将人力嵌入组织中，才能从AI中获得正向投资回报？如果AI如此强大，只需提供即可让人许可使用并从中获益，那为何要让自己陷入资源极其稀缺的困境？”

仅靠自动化难以实现ROI的事实，可从福特等企业的发展中看出。据报道，福特在依赖自动化后，正在重新雇佣老员工。
福特的转变是另一例证，说明AI的市场策略已从“一切皆可自动化”转向“人类不可或缺”。

“依我判断，他们正在承认，AI部署的复杂性远超我们之前被引导相信的程度，”尼科尔森说。“相关成本更高，而ROI的实现将需要更长时间。”

对企业而言，最好明确其AI愿景，再决定如何实现——无论是使用微软的服务、选择其他云服务商，还是结合不同供应商和服务商的工具。
“从根本上说，企业需要理解他们想通过AI实现什么目标、如何实现，然后从目标倒推规划路径，”苏总结道。

英文来源：

Sponsored by Google Cloud
Choosing Your First Generative AI Use Cases
To get started with generative AI, first focus on areas that can improve human experiences with information.
Microsoft Frontier Company is the latest example of how experts are necessary to achieving returns on AI investments.
Microsoft introduced a new operating business on Thursday as it seeks to provide enterprises with the expertise needed to deploy and integrate AI.
The company is investing $2.5 billion into Microsoft Frontier Company, an effort that will embed 6,000 industry and engineering experts to codesign, co-innovate, deploy and improve AI systems with customers. While the program is meant to accelerate AI initiatives, it’s also recognition that AI integration cannot succeed without humans in the enterprise.
While Microsoft said its more than an investment in forward deployed engineering, the new initiative closely follows in the footsteps of organizations such as Palantir, which has championed embedding software engineers into enterprises, as well as AWS’s $1 billion FDE investment earlier this week. OpenAI also introduced a similar program in February.
The push by cloud providers to embed human expertise within customer environments recognizes that, despite the rapid growth in the AI market, gaps exist that prevent enterprises from fully scaling and deploying AI projects.
“A lot of enterprises still struggle with AI adoption,” said Lian Jye Su, an analyst at Omdia, a division of Informa TechTarget. “Even those that adopt it struggle to scale, and those who scale struggle to either maintain it or operate it safely.”
To help close the gap, cloud providers are now providing problem-solving tools Including FDEs.
“What we are seeing in a broader sense is that every single company that used to play in a particular stack within the AI ecosystem is either going up or downstream,” Su added. He pointed to frontier model providers now offering SaaS services and infrastructure, and cloud providers offering integration services and consulting work as additional examples.
The push also suggests vendors are iterating on an earlier stance that AI does not require human helpers.
“All of these AI companies, everyone involved with AI today, has to answer this question: Why are we now being told that you need to embed bodies in organizations so that they can get positive ROI out of AI?” said David Nicholson, an analyst at Futurum Group. “If AI is so good that all you have to do is make it available, and people will license it and use it and get value from it, why would you put yourself in a situation where now you're going to have to scramble to try to find an incredibly scarce resource.”
The fact that ROI is hard to achieve with automation alone can be seen developments from enterprises such as Ford, which is reportedly rehiring legacy employees after relying on automation.
Ford’s reversal is another example that the go-to-market strategy for AI has shifted from "everything can be automated" to "there is a need for people."
“In my assessment, they’re admitting that this is a lot more complicated to deploy than we were led to believe,” Nicholson said. “The cost associated is higher, and the ROI is going to take longer.”
For enterprises, it is best to be clear about their vision for AI and then decide how to get there, whether they will use services from Microsoft, go with another cloud provider or combine tools from different vendors and service providers.
“Fundamentally, enterprises need to understand what they are trying to achieve with AI, how they have to achieve it, and then kind of go backward from there,” Su said.