内容来源：https://www.geekwire.com/2025/microsoft-exec-hints-at-new-copilot-features-saying-ai-will-become-even-more-of-a-team-sport/

内容总结：

微软公司近日表示，正积极推进"人-人-人工智能"协同办公模式的深化应用。该公司人工智能业务高管贾里德·斯帕塔罗在微软校友网络Connect 2025大会上透露，今秋将大规模推出支持人类与AI实时协同工作的新功能，标志着AI应用正从"单人模式"转向"团队协作模式"。

斯帕塔罗指出，AI技术正在重塑职场生态：一方面显著降低专业门槛，使跨岗位协作成为可能；另一方面推动企业重构人才培养体系，资深员工通过"师徒制"帮助新人掌握AI工具使用技巧。他特别提到，领英平台已出现设计师借助AI工具编写代码的新趋势。

尽管个人层面的AI应用价值难以量化，但企业级应用已显现巨大效益。斯帕塔罗坦言，虽然最初对365 Copilot每用户30美元的月费充满信心，但后来发现其真正价值体现在组织层面的流程优化和成本节约，某些案例中甚至实现数亿美元运营成本降低。

值得关注的是，基于GPT-5的Copilot已能胜任专业级创意工作。斯帕塔罗团队在与哈佛大学的研究项目中，AI不仅完成文字润色工作，更成为解决学术争议的"思维伙伴"，其产出质量堪比专业机构。

微软预见的未来组织形态将是"人类主导+智能体运营"模式，疫情时期的数字化转型为当前AI发展奠定了数据基础。斯帕塔罗强调，虽然AI能力突飞猛进，人类仍在战略决策层面发挥着不可替代的核心作用。

中文翻译：

华盛顿州雷德蒙德讯——微软一位高层人工智能主管本周表示，该公司正计划深化"人类与人类及AI的三方协作"，预示着将大规模推广让人与AI协同工作的新功能。微软"AI办公"首席营销官贾里德·斯帕塔罗周二在公司总部会议上表示："迄今为止，基于AI的工作更像是单人运动，今年秋季将明显转变为团队运动——人类将与其他人以及AI协同工作。"

微软去年推出的Copilot Pages功能，已在Microsoft 365中为"多人AI协作"提供了持久化操作界面。斯帕塔罗的发言表明公司正计划拓展这一愿景，并指出OpenAI等合作伙伴也将参与其中。目前微软面临激烈竞争，谷歌、Slack、Zoom及众多AI初创公司都在推出供团队协作使用的聊天机器人和智能体。

斯帕塔罗认为这类功能可能推动Copilot人工智能生产力平台实现病毒式增长，就像疫情期间微软Teams因会议邀请功能而快速普及那样。这些观点是他在"微软校友网络连接2025大会"商业AI趋势论坛上发表的，该论坛由Adobe技术服务公司AI大使兼变革经理比尔·克斯特主持。

斯帕塔罗的其他核心观点包括：
人工智能正日益模糊不同职种的界限。随着AI智能体将"专业知识的边际成本"降为零，人们将能跨角色处理以往被视为专业领域的工作。他以"全栈开发者"为例指出，过去软件开发需要设计、前端、后端等不同专家，而如今领英平台已出现设计师借助AI工具编写大量代码的现象。他认为现代经济长期将人禁锢在狭窄角色中，AI或许能打破这种自亚当·斯密提出劳动分工以来的趋势，释放被过度专业化束缚的人类创造力。

AI正在改变企业的职业发展理念。斯帕塔罗注意到经验丰富的专业人士更能通过Copilot类工具判断AI输出质量，因此部分企业正重启学徒制，让新老员工结对培养AI协作能力。

AI价值在个人层面难以量化。他原以为每月30美元的Microsoft 365 Copilot定价毋庸置疑，后来发现虽然常带来"顿悟时刻"，但仅凭个人生产力提升难以证明成本合理性。真正的投资回报体现在组织层面——某些案例中企业通过流程改进节省了数亿美元运营成本。

AI开始媲美专业工作质量。斯帕塔罗的团队借助GPT-5驱动的Copilot完成哈佛合作项目时，AI不仅处理常规任务，更成为真正的思想伙伴：整合概念、生成精炼文本、化解语言分歧。他感叹"首次感觉外包给专业机构也不会做得更好"，强调AI并非取代人类，而是增强团队能力。

微软预见AI时代将出现新型组织。斯帕塔罗描述的"前沿企业"以人类为主导、智能体执行为特征，人类把握方向而AI承担大量工作。微软重点关注三种模式：人类与助手协作、人类与智能体组队、智能体管理其他智能体。他认为"人类仍扮演至关重要角色，但主要承担领导职能，尽可能由智能体执行操作"。

疫情为当今AI繁荣奠定基础。斯帕塔罗指出远程办公迫使企业实现沟通协作数字化，为AI系统积累了数据基础和行为习惯，"没有昨天的疫情，就不会有今天AI世界的发展"。

GeekWire是本次大会的媒体合作伙伴。

英文来源：

REDMOND, Wash. — Microsoft is preparing to go deeper into “human-to-human-to-AI collaboration,” one of the company’s top AI executives said this week — signaling plans for a broader rollout of features to let people work with each other and AI at the same time.
“So far, AI-based work has been kind of a solo sport, and this fall it will clearly become a team sport, where you’ll be working together with other people and AI,” said Jared Spataro, Microsoft’s chief marketing officer for AI at Work, during a conference Tuesday at the company’s headquarters.
Microsoft introduced Copilot Pages last year as a persistent canvas for “multiplayer AI collaboration” in Microsoft 365. Spataro’s comments indicate that the company is preparing to expand on that vision. He noted that partners such as OpenAI will also play a role.
The company is facing a strong field of competitors, with others such as Google, Slack, Zoom, and AI startups similarly rolling out chatbots and agents for teams to use collaboratively.
Spataro said these kinds of features could create a new degree of viral adoption for Copilot, the company’s AI productivity platform, similar to how Microsoft Teams grew quickly during the pandemic when people began sending each other meeting invites.
His comments were part of a session about trends in AI for business Tuesday at the Microsoft Alumni Network Connect 2025 conference. The discussion with Spataro was led by Bill Kirst, an AI ambassador and change manager for Adobe Technology Services.
More takeaways from Spataro’s remarks:
AI is increasingly blurring the lines between different kinds of jobs.
With AI agents lowering what Spataro called “the marginal cost of expertise” to zero, people will increasingly be able to handle tasks across roles that were previously considered specialties.
He pointed to the “full-stack developer” as a model. In the past, building software often required separate specialists for design, front-end, and back-end work.
Spataro noted that LinkedIn, the Microsoft-owned social network, is already seeing this shift, with some designers writing as much code as developers thanks to AI tools.
Up to this point, he observed, the modern economy has put people into narrow roles that limit their broader impact. AI could free individuals to contribute more widely, reversing a trend dating back to the division of labor described by Adam Smith, the 18th-century economist often called the father of modern economics.
“I think that’s pretty darn cool,” Spataro said. “From our perspective, that would be the unleashing of human ingenuity that somewhat has been locked up because the economy has become uber-specialized.”
AI is changing how companies think about career development.
Spataro noted that experienced professionals tend to get more out of tools like Copilot because they can better judge the quality of AI outputs. Recognizing this, he said, some companies are revisiting apprenticeship programs — pairing experienced employees with newer workers to help them build the judgment and skills to work effectively with AI tools.
AI’s value can be hard to quantify at the individual level.
Spataro recalled thinking early on that Microsoft 365 Copilot would be a no-brainer at $30 per user per month, but he later realized that, while people see plenty of “aha” moments, the cost can be much tougher to justify based on personal productivity alone.
“I thought, holy smokes, this thing, who’s not going to want to pay $30 a month for this?” he said. “Boy, did I learn a lesson.”
He said the real ROI emerges at the organizational level, where companies can measure process improvements and cost savings — in some cases reducing operating expenses by hundreds of millions of dollars.
AI is starting to rival professional work quality.
Spataro described how his own team recently turned to Copilot, powered by GPT-5, when they were struggling to frame ideas for a research project with Harvard. Instead of handling routine tasks, the AI acted as a true thought partner — combining concepts, generating polished prose, and helping resolve disagreements over language.
“For the first time, we felt like we could have gone out to an agency and we wouldn’t have gotten better,” he said. “So that was one that was very inspiring to us. It wasn’t that it was replacing us or the writers or some of the research leads on my team, but it was definitely making us better.”
Microsoft sees a new kind of organization emerging in the AI era.
Spataro described these “frontier firms” as human-led, agent-operated — with people setting direction while AI agents carry out much of the work. He outlined three patterns Microsoft is watching closely: humans working with an assistant, humans teaming with agents, and processes where agents manage other agents.
“Humans still play, we think, an incredibly important role, but they do the leading, and then, as much as possible, agents are operating a lot of what happens,” he said.
The pandemic set the stage for today’s AI boom.
Spataro said the sudden shift to remote and hybrid work forced companies to digitize communication and collaboration, creating the data and habits that AI systems now build upon. “We don’t believe that we could have the world of AI that we have today without the pandemic that we had yesterday,” he said.
GeekWire is a media partner of the Microsoft Alumni Network Connect 2025 conference.